AI Generator Canva

AI Generator Canva — independent reviews, comparisons, pricing and step-by-step guides on Aizhi.

Rabbit r1

The Rabbit r1 is an artificial intelligence personal assistant device developed by the American technology startup Rabbit Inc and co-designed by Teenage Engineering. It was announced at the 2024 Consumer Electronics Show as a handheld device intended to perform digital tasks through voice commands, touch interaction, and web-based AI agents. The r1 was marketed around Rabbit's concept of a "large action model" (LAM), which the company described as software able to operate websites and services on behalf of users. The device runs rabbitOS, an operating system based on the Android Open Source Project. Its services have included AI search, image recognition, voice interaction, music playback, rideshare and food-ordering integrations, and later experimental web-agent features such as LAM Playground and teach mode. Initial reviews were largely negative, with reviewers criticizing the device's limited functionality, bugs, and unclear advantages over a smartphone. Critics also questioned Rabbit's claims after the r1 software was shown to run on an Android phone. Rabbit continued to issue software updates after launch, including rabbitOS 2 in September 2025, which introduced a redesigned card-based interface, gesture navigation, and a "creations" feature for generating small software tools and experiences on the device. Rabbit Inc was founded by Jesse Lyu Cheng. == Hardware == Display: A 2.88-inch touchscreen for interactive user input. Input: push-to-talk button to activate voice commands; scroll wheel; Gyroscope; Magnetometer; Accelerometer; GPS. Camera: 8 MP single camera, with a resolution of 3264x2448, allowing for the connected external AI to use computer vision. Audio: Equipped with a speaker and dual microphones for audio interaction. Connectivity: Supports Wi-Fi and cellular connections via a SIM card slot to access internet services. Processor: Runs on a 2.3GHz MediaTek Helio P35 processor. Memory: Contains 4GB of RAM for operational tasks. Storage: Offers 128GB of internal storage for data. Ports: Utilizes a USB-C port for charging and data connections. == Software == The Rabbit r1 runs rabbitOS, which is based on the Android Open Source Project (AOSP), specifically Android 13. Rabbit founder Jesse Lyu described rabbitOS as a "very bespoke AOSP" after reports that the r1's software could be run on a conventional Android phone. Rabbit described the r1 as using a large action model (LAM), a type of AI agent intended to perform tasks across software interfaces rather than only answer questions. At launch, the device supported a limited set of services, including AI search, vision features, music playback, and some third-party integrations. Perplexity.ai was one of the AI services used to answer user queries. In 2024, Rabbit released several software updates that added features and attempted to address early criticism of the device. In July 2024, the company launched "beta rabbit", an advanced search and conversation mode for more complex queries. In October 2024, it released LAM Playground, a web-based agent feature intended to let the r1 operate websites on behalf of users. Reviewers found the feature experimental; Android Authority reported that it could perform some navigation tasks but struggled with CAPTCHAs, loops, and unintended behavior. In November 2024, Rabbit introduced a beta "teach mode", which allowed users to demonstrate web-based tasks in the Rabbithole web portal and later ask the r1 to repeat them. The company described teach mode as experimental, and The Verge noted that Rabbit warned users that results could be unpredictable and that CAPTCHA-protected sites could cause problems. Rabbit released rabbitOS 2 in September 2025. The update redesigned the interface around a card-based layout, added additional touchscreen gestures, and introduced "creations", a feature that lets users generate simple software tools, games, and interfaces through natural-language prompts. Coverage of the update described it as a major software overhaul rather than new hardware. == Reception == === Funding === Rabbit raised $20 million in funding from Khosla Ventures, Synergis Capital and Kakao Investment in October 2023. The company announced an additional $10 million in funding in December 2023. === Sales === Following its announcement at the 2024 Consumer Electronics Show, 130,000 units were sold. On August 13, 2024, Rabbit announced that sales of r1 had expanded to the entire European Union (except Malta) and United Kingdom. On August 21, 2024, sales of r1 expanded to Singapore. === Reviews === The r1 was met with strong criticism immediately after Rabbit began shipping the device. Some reviews questioned what the device was able to do that a smartphone could not, while comparing it to the similar Humane Ai Pin. YouTuber Marques Brownlee called the device "barely reviewable". Android Authority's Mishaal Rahman managed to install Rabbit r1's software on a Pixel 6a smartphone, after a tipster shared an APK file. The Verge echoed the claims made by Rahman. In response, Lyu published statements confirming its use of Android, but denying that the r1 is an Android app. Mashable called its Vision features impressive, but said that "these praise-worthy features are overshadowed by buggy performance". Ars Technica wrote a blog post claiming "the company is blocking access from bootleg APKs". TechCrunch gave a slightly more positive review, calling the device a "fun peep at a possible future", but could not "advise anyone to buy one now." Shortly after the launch of r1, Rabbit began a weekly cadence of software updates to address much of the criticism from the early reviews, including "battery and GPS performance, time zone selection, and more". Digital Trends said the Magic Camera feature "takes the most mundane, ordinary, and badly composed photos and makes something fun and eye-catching from them." Mashable said the "beta rabbit" feature "makes Rabbit R1 more conversational and intelligent". Later coverage noted that Rabbit continued to update the r1 after its poorly received launch. The Verge reported in September 2024 that about 5,000 of roughly 100,000 purchasers were using the device at any given moment, citing Lyu, and described the product as having launched before it was ready. In 2025, coverage of rabbitOS 2 described the update as an attempt to reset the device's software experience after the criticism of its original release. == Controversies == === GAMA project === Rabbit Inc has garnered attention due to allegations surrounding its funding and the company's past projects. The company came under scrutiny when Stephen Findeisen, known as Coffeezilla on YouTube, published a video in May 2024, alleging that Rabbit Incorporation was "built on a scam". Rabbit Incorporation, initially named Cyber Manufacturing Co, rebranded just two months before launching the Rabbit R1. The company, under its former name, raised $6 million in November 2021 for a project called GAMA, described as a "Next Generation NFT Project." Jesse Lyu, the CEO of Rabbit Incorporation, referred to GAMA as a "fun little project." Coffeezilla, who investigates influencer scams, highlighted old Clubhouse recordings of Jesse Lyu discussing the GAMA project. In these recordings, Lyu emphasized the substantial funding behind GAMA and its potential to be a revolutionary, carbon-negative cryptocurrency. Coffeezilla questioned the whereabouts of the funds raised for GAMA, estimating that approximately $1 million in refunds to investors remained unresolved. He suggested that the rebranding to Rabbit Incorporation and the shift to developing the Rabbit R1 were attempts to divert from the GAMA project's issues. In response to Coffeezilla's inquiries, Rabbit Incorporation stated that the $6 million raised was used for the GAMA project. The company said that NFTs cannot be refunded unless the owner agrees to "burn" them on the blockchain. Rabbit Incorporation also said that the GAMA project was open-sourced and returned to the community, aligning with community feedback. They also mentioned that efforts to buy back NFTs were made to counteract malicious trading and maintain market stability. === Security === In June 2024, Engadget reported that the Rabbitude team, a community reverse engineering project, had gained access to the r1's codebase revealing that r1's software contained several hardcoded API keys in its code for ElevenLabs, Microsoft Azure, Yelp, and Google Maps, potentially allowing unauthorized access to r1 responses, including those containing the users' personal information. For a short time, Rabbit immediately began revoking and rotating those secrets and confirmed that the code was leaked by an employee who had "been terminated and remains under investigation". In July 2024, the company revealed that all user chats and device pairing data were logged on the r1 with no ability to delete them. This meant that lost or stolen devices could be used to extract user
Read more →
Course of Action Display and Evaluation Tool

Course of Action Display and Evaluation Tool (CADET) was a research program, and the eponymous prototype software system, that applied knowledge-based techniques of Artificial Intelligence to the problem of battle planning. CADET was also known as Course of Action Display and Elaboration Tool. It was considered an early example of such systems and was funded by the United States Army and by the Defense Advanced Research Projects Agency (DARPA). CADET influenced a later DARPA program called RAID which in turn produced a technology adopted by the United States Army and the United States Marine Corps. == History == The development of Course of Action Display and Evaluation Tool (CADET) began in 1996, at the Carnegie Group, Inc., Pittsburgh PA, funded under the Small Business Innovation Research (SBIR) program. The goal of the first phase SBIR project was to produce “...a live storyboard of [Course of Action] COA development, wargaming, animation, and assessment.” In 1997, the United States Army awarded the Carnegie Group Inc. $750K for SBIR Phase II. The intent was to develop “...a war-gaming modeling and analysis Decision Support System (DSS), … CADET will consist of a combination of Knowledge-Based and decision analytic tools and technologies to provide fast nimble COA war-gaming modeling, simulation, and animation under direct control of the commander and staff. ...Phase II will result in an operations prototype (OP) suitable for use and evaluation in field exercises.” In 2000, CADET was integrated and experimentally evaluated within the framework of the Integrated Course of Action Critiquing and Elaboration System (ICCES) experiment, conducted by the Battle Command Battle Laboratory – Leavenworth (BCBL-L) within the program Concept Experimentation Program (CEP) sponsored by TRADOC. In 2000-2002, DARPA applied CADET in the program titled Command Post of the Future (CPoF) as a tool to generate a course of action. Under the umbrella of the CPoF program, CADET was integrated with the FOX GA system to provide a detailed planner, coupled with COA generation capability. In the same period, Battle Command Battle Lab-Huachuca (BCBL-H) performed an integration CADET with the system called All Source Analysis System-Light (ASAS-L); here CADET was intended to generate plans for intelligence assets, and conduct wargames of different COAs, enemy versus friendly. From 1996 through 2002, work on CADET was performed by the Carnegie Group, Inc., and supported by funding from the US Army CECOM (CADET SBIR Phase I, CADET SBIR Phase II and CADET Enhancements); DARPA (Command Post of the Future); and TRADOC BCBL-H. == Operation == CADET was intended to be used by the staff of the United States Army Brigade, within the Military Decision Making Process (MDMP). In particular, CADET helped produce, automatically or semi-automatically, the products generated within the step of MDMP called Course of Action (COA) Development and the following step of MDMP called COA Analysis and Wargaming. CADET software resided on a laptop computer. Using the computer, the staff officers entered the input to CADET, or alternatively this input arrived at CADET from upstream computer systems. The input consisted of: Order of Battle, i.e., the units constituting the friendly brigade and the enemy units participating in the battle, and their various characteristics; primary activities of the Course of Action, where each activity is typically linked to one or more geographic areas or a route, and sometimes to a major unit executing the activity; digital map of the region where the battle was to take place, including the digital description of significant features such as locations of friendly and enemy units, roads, assembly areas, objectives, and axes of attacks. Taking this input, CADET automatically performed the following tasks (not sequentially): Planning and scheduling the low-level tasks necessary for a given COA Allocating tasks to various units and assets constituting the brigade Assigning suitable locations and routes Estimating the battle losses (attrition) of friendly and enemy forces, and consumption of resources (e.g., fuel and ammunition) Predicting enemy actions or reactions. CADET produced the following outputs: Synchronization matrix, directly editable and printable; synchronization matrix is a kind of Gantt chart that shows assignments of activities to units, to locations/routes and to time periods Map overlays in PPT or JPG formats Animation output XML formally-encoded plan Textual Operation Plan (OPLAN) draft E-mail messages with attachments: XML and text versions of OPLAN == Design == The core algorithm is a planning algorithm where CADET uses a knowledge-based approach of the hierarchical-task-network type. Each task class is associated with a model of more detailed subtasks that should be performed in order to accomplish the higher-level task. Algorithms selected (heuristically) a task and then decomposes it into subtasks. Although similar to hierarchical-task-network planning algorithm, CADET’s algorithm includes elements of adversarial reasoning. After adding a subtask, the algorithm uses rules to determine the enemy’s probable actions and reactions as well as friendly counteractions This approximated the action-reaction-counteraction technique of manual wargaming used by the United States Army. When a task involves movements of a unit, the algorithm performs routing, i.e., finds a route for the movement that minimizes the time required for the movement as well as exposure to the enemy attacks. Each added tasks (subtask) normally requires a unit which would execute the task, and a time period when the task would be executed. Therefore, when a certain number of subtasks is added by the planning process, the algorithm also performs the allocation of the newly added subtasks to units and to time periods (i.e., scheduling). allocation and scheduling of tasks relies on both domain-specific and constraint-guided heuristics. A tasks may also require expenditures of fuel and ammunition. If the tasks involves engagement with the enemy, the performing units will experience lossesof personnel and weapon systems (attrition). CADET’s algorithm includes estimates of consumption of different types of consumables, and also attrition. Depending on the degree of attrition and consumption, CADET adds tasks that are needed to refuel or reconstitute the units. The algorithm continually interleaves incremental steps of planning, routing, scheduling, and attrition and consumption estimates. == Evaluation == Two evaluation experiments are described in literature. The first experiment called ICCES took three days. The subjects were Army officers from combat arms branches, with 11 to 23 years of active service, in the ranks of majors and lieutenant colonels, a total of 8. Each officer was given 4 hours of training learning to operate CADET and related computer tools. Officers were divided into two groups and given a tactical scenario. One group (the control group) used the traditional, manual process; the other used the system called ICCES, the automated core of which was CADET. Each group produced three COA sketches and statements and one COA synchronization matrix. Then, the experiment was repeated with another scenario but the control group became the automated group and vice versa. The users were generally satisfied with the quality of the ICCES-generated products. The group using ICCES made only a few changes to the product that was automatically generated, indicating that they agreed with the majority of the plan that ICCES produced. The second experiment was reminiscent of Turing test. The experiment involved one user, nine judges (active-duty officers, mainly colonels and lieutenant colonels), and five scenarios obtained from several US Army exercises. For each scenario, experimenters obtained synchronization matrices that were produced in earlier exercises, typically by a team of four to five officers in three to four hours, spending approximately 16 person-hours in total. Using these scenarios and COAs, the user had CADET generate automatically detailed plans and express them as synchronization matrices. The user, a retired US Army officer, reviewed and slightly edited the matrices. The entire process took less than two minutes of computations by and approximately 20 minutes of review and post-editing, approximately 0.4 person-hour in total per product. The experimenters gave the resulting matrices the same visual style as those produced by humans. The judges, who did not know whether a planning product was a traditional product of humans, or with computerized aids, were asked to grade the products. The result was that the average grades for manual products and CADET-generated products were statistically indistinguishable, even though CADET-generated products required far less time to produce. == Legacy == CADET served as “...an example of how even relatively basic A
Read more →
Model Context Protocol

The Model Context Protocol (MCP) is an open standard and open-source framework introduced by Anthropic in November 2024 to standardize the way artificial intelligence (AI) systems like large language models (LLMs) integrate and share data with external tools, systems, and data sources. MCP provides a standardized interface for reading files, executing functions, and handling contextual prompts. Following its announcement, the protocol was adopted by major AI providers, including OpenAI and Google DeepMind. == Background == MCP was announced by Anthropic in November 2024 as an open standard for connecting AI assistants to data systems such as content repositories, business management tools, and development environments. The protocol was created at Anthropic by engineers David Soria Parra and Justin Spahr-Summers. It aims to address the challenge of information silos and legacy systems. Before MCP, developers often had to build custom connectors for each data source or tool, resulting in what Anthropic described as an "N×M" data integration problem. Earlier stop-gap approaches—such as OpenAI's 2023 "function-calling" API and the ChatGPT plug-in framework—solved similar problems but required vendor-specific connectors. MCP re-uses the message-flow ideas of the Language Server Protocol (LSP) and is transported over JSON-RPC 2.0. In December 2025, Anthropic donated the MCP to the Agentic AI Foundation (AAIF), a directed fund under the Linux Foundation, co-founded by Anthropic, Block and OpenAI, with support from other companies. == Features == The protocol was released with software development kits (SDKs) in programming languages including Python, TypeScript, C# and Java. Anthropic maintains an open-source repository of reference MCP server implementations and SDKs. MCP defines a standardized framework for integrating AI systems with external data sources and tools. It includes specifications for data ingestion and transformation, contextual metadata tagging, and AI interoperability across different platforms. The protocol also supports bidirectional connections between data sources and AI tools. MCP enables applications such as querying structured databases with plain language in the field of natural language data access. The protocol is used in AI-assisted software development tools. Integrated development environments (IDEs), coding platforms such as Replit, and code intelligence tools like Sourcegraph have adopted MCP to grant AI coding assistants real-time access to project context. MCP Apps is an official extension to the Model Context Protocol built on mcp-ui. While the base MCP specification is restricted to text and structured data, MCP Apps standardizes the delivery of interactive user interfaces—such as dashboards, forms, and data visualizations—from MCP servers to host applications like Claude and ChatGPT. == Adoption == In March 2025, OpenAI officially adopted the MCP, after having integrated the standard across its products, including the ChatGPT desktop app. In September 2025, OpenAI added support for MCP to ChatGPT apps. This allows for third-party access inside ChatGPT. MCP can be integrated with Microsoft Semantic Kernel, and Azure OpenAI. MCP servers can be deployed to Cloudflare. In April 2026, the AAIF held the MCP Dev Summit North America in New York City, drawing approximately 1,200 attendees. == Reception == The Verge reported that MCP addresses a growing demand for AI agents that are contextually aware and capable of pulling from diverse sources. In April 2025, security researchers released an analysis that concluded there are multiple outstanding security issues with MCP, including prompt injection, tool permissions that allow for combining tools to exfiltrate data, and lookalike tools that can silently replace trusted ones. MCP has been likened to OpenAPI, a similar specification that aims to describe APIs.
Read more →
Noise-based logic

Noise-based logic (NBL) is a class of multivalued deterministic logic schemes, developed in the twenty-first century, where the logic values and bits are represented by different realizations of a stochastic process. The concept of noise-based logic and its name was created by Laszlo B. Kish. In its foundation paper it is noted that the idea was inspired by the stochasticity of brain signals and by the unconventional noise-based communication schemes, such as the Kish cypher. == The noise-based logic space and hyperspace == The logic values are represented by multi-dimensional "vectors" (orthogonal functions) and their superposition, where the orthogonal basis vectors are independent noises. By the proper combination (products or set-theoretical products) of basis-noises, which are called noise-bit, a logic hyperspace can be constructed with D(N) = 2N number of dimensions, where N is the number of noise-bits. Thus N noise-bits in a single wire correspond to a system of 2N classical bits that can express 22N different logic values. Independent realizations of a stochastic process of zero mean have zero cross-correlation with each other and with other stochastic processes of zero mean. Thus the basis noise vectors are orthogonal not only to each other but they and all the noise-based logic states (superpositions) are orthogonal also to any background noises in the hardware. Therefore, the noise-based logic concept is robust against background noises, which is a property that can potentially offer a high energy-efficiency. == The types of signals used in noise-based logic == In the paper, where noise-based logic was first introduced, generic stochastic-processes with zero mean were proposed and a system of orthogonal sinusoidal signals were also proposed as a deterministic-signal version of the logic system. The mathematical analysis about statistical errors and signal energy was limited to the cases of Gaussian noises and superpositions as logic signals in the basic logic space and their products and superpositions of their products in the logic hyperspace (see also. In the subsequent brain logic scheme, the logic signals were (similarly to neural signals) unipolar spike sequences generated by a Poisson process, and set-theoretical unifications (superpositions) and intersections (products) of different spike sequences. Later, in the instantaneous noise-based logic schemes and computation works, random telegraph waves (periodic time, bipolar, with fixed absolute value of amplitude) were also utilized as one of the simplest stochastic processes available for NBL. With choosing unit amplitude and symmetric probabilities, the resulting random-telegraph wave has 0.5 probability to be in the +1 or in the −1 state which is held over the whole clock period. == The noise-based logic gates == Noise-based logic gates can be classified according to the method the input identifies the logic value at the input. The first gates analyzed the statistical correlations between the input signal and the reference noises. The advantage of these is the robustness against background noise. The disadvantage is the slow speed and higher hardware complexity. The instantaneous logic gates are fast, they have low complexity but they are not robust against background noises. With either neural spike type signals or with bipolar random-telegraph waves of unity absolute amplitude, and randomness only in the sign of the amplitude offer very simple instantaneous logic gates. Then linear or analog devices unnecessary and the scheme can operate in the digital domain. However, whenever instantaneous logic must be interfaced with classical logic schemes, the interface must use correlator-based logic gates for an error-free signal. == Universality of noise-based logic == All the noise-based logic schemes listed above have been proven universal. The papers typically produce the NOT and the AND gates to prove universality, because having both of them is a satisfactory condition for the universality of a Boolean logic. == Computation by noise-based logic == The string verification work over a slow communication channel shows a powerful computing application where the methods is inherently based on calculating the hash function. The scheme is based on random telegraph waves and it is mentioned in the paper that the authors intuitively conclude that the intelligence of the brain is using similar operations to make a reasonably good decision based on a limited amount of information. The superposition of the first D(N) = 2N integer numbers can be produced with only 2N operations, which the authors call "Achilles ankle operation" in the paper. == Computer chip realization of noise-based logic == Preliminary schemes have already been published to utilize noise-based logic in practical computers. However, it is obvious from these papers that this young field has yet a long way to go before it will be seen in everyday applications.
Read more →
Language identification

In natural language processing, language identification or language guessing is the problem of determining which natural language a given content is in. Computational approaches to this problem view it as a special case of text categorization, solved with various statistical methods. == Overview == === Logical approach === A common non-statistical intuitive approach (though highly uncertain) is to look for common letter combinations, or distinctive diacritics or punctuation. === Statistical approach === There are several statistical approaches to language identification. An older statistical method by Grefenstette was based on the frequency of short n-grams, which are often function morphemes. For example, "ing" is more common in English than in French, while the sequence "que" is more common in French. Given a new page found on the Web, one counts the number of occurrences of each such short sequence and picks the language whose frequency table it matches the most. One technique is to compare the compressibility of the text to the compressibility of texts in a set of known languages. This approach is known as mutual information based distance measure. The same technique can also be used to empirically construct family trees of languages which closely correspond to the trees constructed using historical methods. Mutual information based distance measure is essentially equivalent to more conventional model-based methods and is not generally considered to be either novel or better than simpler techniques. Another technique, as described by Cavnar and Trenkle (1994) and Dunning (1994) is to create a language n-gram model from a "training text" for each of the languages. These models can be based on characters (Cavnar and Trenkle) or encoded bytes (Dunning); in the latter, language identification and character encoding detection are integrated. Then, for any piece of text needing to be identified, a similar model is made, and that model is compared to each stored language model. The most likely language is the one with the model that is most similar to the model from the text needing to be identified. This approach can be problematic when the input text is in a language for which there is no model. In that case, the method may return another, "most similar" language as its result. Also problematic for any approach are pieces of input text that are composed of several languages, as is common on the Web. As of 2025, a commonly used baseline method is via the fastText library, which has comparable classification accuracy as deep learning techniques, but much faster. == Identifying similar languages == One of the great bottlenecks of language identification systems is to distinguish between closely related languages. Similar languages like Bulgarian and Macedonian or Indonesian and Malay present significant lexical and structural overlap, making it challenging for systems to discriminate between them. In 2014 the DSL shared task has been organized providing a dataset (Tan et al., 2014) containing 13 different languages (and language varieties) in six language groups: Group A (Bosnian, Croatian, Serbian), Group B (Indonesian, Malaysian), Group C (Czech, Slovak), Group D (Brazilian Portuguese, European Portuguese), Group E (Peninsular Spanish, Argentine Spanish), Group F (American English, British English). The best system reached performance of over 95% results (Goutte et al., 2014). Results of the DSL shared task are described in Zampieri et al. 2014. == Software == Apache OpenNLP includes char n-gram based statistical detector and comes with a model that can distinguish 103 languages Apache Tika contains a language detector for 18 languages
Read more →
Interim Measures for the Management of Generative AI Services

The Interim Measures for the Management of Generative AI Services (Chinese: 生成式人工智能服务管理暂行办法; pinyin: Shēngchéng shì réngōng zhìnéng fúwù guǎnlǐ zànxíng bànfǎ) are a set of regulations governing public-facing generative artificial intelligence services in China. Issued on 10 July 2023 and effective from 15 August 2023, they were China's first binding regulation specifically targeting generative AI. They have been described as among the earliest such regulations adopted by any country. The measures were jointly issued by the Cyberspace Administration of China (CAC) and six other national bodies: the National Development and Reform Commission, the Ministry of Education, the Ministry of Science and Technology, the Ministry of Industry and Information Technology, the Ministry of Public Security, and the National Radio and Television Administration. Among the measures' most prominent requirements is that generative AI services must uphold Core Socialist Values and must not generate content that could subvert state power, harm national security, or undermine social stability. The measures also require providers of public-facing generative AI services to undergo security assessments and register their algorithms with the CAC. As of December 2025, 748 generative AI services had completed the filing process at the national level. == Background == The Interim Measures build on two earlier sets of regulations targeting specific algorithm applications. The Administrative Provisions on Algorithm Recommendation for Internet Information Services, effective from March 2022, established China's algorithm registry and required providers of recommendation algorithms with "public opinion properties or social mobilization capabilities" to file with the CAC and undergo security assessments. The Administrative Provisions on Deep Synthesis of Internet Information Services, effective from January 2023, extended similar requirements to algorithms used for generating synthetic media such as deepfakes. In April 2023, the CAC released a draft of the generative AI regulation for public comment. The draft included several requirements that attracted attention, including that generated content should "embody Core Socialist Values" and that training data should be "true and accurate". The public consultation period ran until May 2023. The final version, published in July 2023, was substantially revised from the draft. According to an analysis by the Future of Privacy Forum, changes appeared to reflect feedback from industry stakeholders including Baidu, Xiaomi, SenseTime, and others, as well as input from government-affiliated research institutes. The final measures adopted a more permissive tone, with the CAC describing its approach as "inclusive and prudent" (包容审慎) and emphasising "classified and graded" (分类分级) supervision. == Scope == The measures apply to services that use generative AI technology to provide text, images, audio, video, or other content to the public within mainland China (Article 2). They do not apply to organisations that develop or use generative AI internally without offering services to the domestic public, such as industry associations, enterprises, and research institutions. Overseas providers whose services are accessible to users in China are also subject to the measures. == Key provisions == === Content requirements === Article 4 sets out the core content obligations. Providers and users of generative AI services must uphold the Core Socialist Values. The measures prohibit generating content that incites subversion of national sovereignty or the socialist system, endangers national security or the nation's image, incites separatism, promotes terrorism or extremism, promotes ethnic hatred or discrimination, or contains violence, obscenity, or false information prohibited by law. These content prohibitions largely mirror those in Article 12 of the Cybersecurity Law and in prior regulations governing online content. Article 4 also requires that models be designed and trained to avoid discrimination, that services respect intellectual property rights, and that providers take effective measures to improve the transparency and accuracy of generated content. === Training data and labelling === Article 7 requires providers to ensure that training data is of high quality and legitimately sourced, and that it does not infringe upon intellectual property rights. Where personal information is used, consent must be obtained. The final version of this provision removed language from the draft that would have held providers responsible for the "legitimacy" of all pretraining data, replacing it with a requirement to "employ effective measures to improve the quality of training data". Article 8 requires providers to establish labelling rules for training data and to conduct quality assessments of data annotations. Article 12 requires that generated images, videos, and other synthetic content be labelled as AI-generated. === User rights and privacy === Article 11 requires providers to protect user privacy, to minimise the collection and retention of personal data, and to refrain from unlawfully sharing user information. Users have the right to request review, correction, or deletion of their personal information. Article 10 requires providers to take measures to prevent excessive dependence on or addiction to generative AI services by minors. === Security assessment and algorithm filing === Article 17 requires that providers of generative AI services with "public opinion properties or the capacity for social mobilization" (具有舆论属性或者社会动员能力) carry out security assessments and complete algorithm filing procedures in accordance with the Administrative Provisions on Algorithm Recommendation for Internet Information Services. == Implementation == === Algorithm filing process === In practice, the filing requirements under the Interim Measures have developed into a two-tier process. The first tier is the standard algorithm filing (算法备案) under the pre-existing Algorithm Recommendation Provisions, which involves submitting information about an algorithm's design, purpose, and data sources to the CAC. This process is primarily a registration mechanism. For public-facing generative AI products, there is an additional, more rigorous process commonly referred to as the "large model filing" (大模型备案). This involves submitting a security self-assessment report, data annotation rules, a keyword blocking list, and evaluation test question sets. The process includes technical testing at the provincial level, followed by review at the national CAC level. The algorithm filing targets specific algorithms, while the large model filing evaluates the broader system architecture, training data, model parameters, and potential social impact. The CAC publishes lists of generative AI services that have successfully completed the filing process. The first such list was published on 2 April 2024. According to the CAC's year-end announcements, 302 generative AI services had completed national-level filing by the end of 2024 (of which 238 were new that year), alongside 105 applications that completed local-level registration. By the end of 2025, the cumulative total had risen to 748 national-level filings and 435 local-level registrations. === Content compliance and testing === According to the Carnegie Endowment, the CAC has conducted compliance audits of generative AI services with a particular focus on ensuring appropriate responses to queries about politically sensitive topics. The large model filing process requires providers to pass both provincial-level and national-level technical testing before their services can be made available to the public. On 1 March 2024, the National Technical Committee 260 on Cybersecurity (TC260) published TC260-003, the Basic Security Requirements for Generative AI Services (生成式人工智能服务安全基本要求), a technical standard that provides detailed guidance on the security assessments required under the Interim Measures. The standard covers requirements for training data safety, model security, and content safety evaluation, and is used as a reference for the filing process. == Analysis == === Relationship to broader Chinese internet regulation === The content requirements in the Interim Measures extend China's existing framework for online information control to generative AI. Legal scholars have noted that the "Core Socialist Values" provision and the specific content prohibitions are consistent with longstanding requirements imposed on internet platforms under the Cybersecurity Law and related regulations. The Asia Society Policy Institute has described the Chinese government's highest regulatory priority in this area as retaining control of information, noting that content-related obligations receive stricter enforcement than other provisions. === Nature of the filing system === The character of the filing system has been debated by scholars. Angela Huyue Zh
Read more →
Generative engine optimization

Generative engine optimization (GEO) is one of the names given to the practice of structuring digital content and managing online presence to improve visibility in responses generated by generative artificial intelligence (AI) systems. The practice influences the way large language models (LLMs) retrieve, summarize, and present information in response to user queries. Related terms include answer engine optimization (AEO) and artificial intelligence optimization (AIO). The concept of GEO first appeared in response to generative AI technologies being integrated into mainstream search and information retrieval systems. Tools are used to monitor how websites and brands are cited, referenced, or incorporated into responses produced by large language models. == Terminology == Several overlapping terms describe related practices, and usage varies across practitioners, vendors, and publications. No consensus definition distinguishing these terms had been established in the academic literature as of early 2026, and the terms are frequently used interchangeably in trade and practitioner contexts. Other terms for the same concept include answer engine optimization (AEO), large language model optimization (LLMO), artificial intelligence optimization (AIO), and AI SEO. In 2026, Google released documentation entitled "Optimizing your website for generative AI features on Google Search." According to this documentation, "optimizing for generative AI search is optimizing for the search experience, and thus still SEO.” This position had previously been shared at conferences, with 2026 being the first time Google released official documentation stating it. == Factors influencing generative engine optimization == By early 2026, the focus of GEO practitioners shifted from simple keyword placement to "semantic relevance", a metric driven by the integration of advertising into conversational AI. OpenAI and Google began monetizing AI search results, which is not currently considered an aspect of generative engine optimization but is adjacent.
Read more →
Combs method

The Combs method is a rule base reduction method of writing fuzzy logic rules described by William E. Combs in 1997. It is designed to prevent combinatorial explosion in fuzzy logic rules. The Combs method takes advantage of the logical equality ( ( p ∧ q ) ⇒ r ) ⟺ ( ( p ⇒ r ) ∨ ( q ⇒ r ) ) {\displaystyle ((p\land q)\Rightarrow r)\iff ((p\Rightarrow r)\lor (q\Rightarrow r))} . == Equality proof == The simplest proof of given equality involves usage of truth tables: == Combinatorial explosion == Suppose we have a fuzzy system that considers N variables at a time, each of which can fit into at least one of S sets. The number of rules necessary to cover all the cases in a traditional fuzzy system is S N {\displaystyle S^{N}} , whereas the Combs method would need only S × N {\displaystyle S\times N} rules. For example, if we have five sets and five variables to consider to produce one output, covering all the cases would require 3125 rules in a traditional system, while the Combs method would require only 25 rules, taming the combinatorial explosion that occurs when more inputs or more sets are added to the system. This article will focus on the Combs method itself. To learn more about the way rules are traditionally formed, see fuzzy logic and fuzzy associative matrix. == Example == Suppose we were designing an artificial personality system that determined how friendly the personality is supposed to be towards a person in a strategic video game. The personality would consider its own fear, trust, and love in the other person. A set of rules in the Combs system might look like this: The table translates to: [IF Fear IS Unafraid THEN Friendship IS Enemies OR IF Fear IS ModerateFear THEN Friendship IS Neutral OR IF Fear IS Afraid THEN Friendship IS GoodFriends ] OR [IF Trust IS Distrusting THEN Friendship IS Enemies OR IF Trust IS ModerateTrust THEN Friendship IS Neutral OR IF Trust IS Trusting THEN Friendship IS GoodFriends] OR [IF Love IS Unloving THEN Friendship IS Enemies OR IF Love IS ModerateLove THEN Friendship IS Neutral OR IF Love IS Loving THEN Friendship IS GoodFriends] In this case, because the table follows a straightforward pattern in the output, it could be rewritten as: Each column of the table maps to the output provided in the last row. To obtain the output of the system, we just average the outputs of each rule for that output. For example, to calculate how much the computer is Enemies with the player, we take the average of how much the computer is Unafraid, Distrusting, and Unloving of the player. When all three averages are obtained, the result can then be defuzzified by any of the traditional means.
Read more →
ALL-IN-1

ALL-IN-1 was an office automation product developed and sold by Digital Equipment Corporation in the 1980s. It was one of the first purchasable off the shelf electronic mail products. It was later known as Office Server V3.2 for OpenVMS Alpha and OpenVMS VAX systems before being discontinued. == Overview == ALL-IN-1 was advertised as an office automation system including functionality in Electronic Messaging, Word Processing and Time Management. It offered an application development platform and customization capabilities that ranged from scripting to code-level integration. ALL-IN-1 was designed and developed by Skip Walter, John Churin and Marty Skinner from Digital Equipment Corporation who began work in 1977. Sheila Chance was hired as the software engineering manager in 1981. The first version of the software, called CP/OSS, the Charlotte Package of Office System Services, named after the location of the developers, was released in May 1982. In 1983, the product was renamed ALL-IN-1 and the Charlotte group continued to develop versions 1.1 through 1.3. Digital then made the decision to move most of the development activity to its central engineering facility in Reading, United Kingdom, where a group there took responsibility for the product from version 2.0 (released in field test in 1984 and to customers in 1985) onward. The Charlotte group continued to work on the Time Management subsystem until version 2.3 and other contributions were made from groups based in Sophia Antipolis, France (System for Customization Management and the integration with VAX Notes), Reading (Message Router and MAILbus), and Nashua, New Hampshire (FMS). ALL-IN-1 V3.0 introduced shared file cabinets and the File Cabinet Server (FCS) to lay the foundation for an eventual integration with TeamLinks, Digital's PC office client. Previous integrations with PCs included PC ALL-IN-1, a DOS-based product introduced in 1989 that never proved popular with customers. Bob Wyman was the first product manager. He oversaw the growth of the product culminating in over $2 billion per year in revenue and market leadership in the proprietary office automation sector. Other consultants from Digital Equipment Corporation involved include Frank Nicodem, Donald Vickers and Tony Redmond.
Read more →
Cooperative coevolution

Cooperative Coevolution (CC) in the field of biological evolution is an evolutionary computation method. It divides a large problem into subcomponents, and solves them independently in order to solve the large problem. The subcomponents are also called species. The subcomponents are implemented as subpopulations and the only interaction between subpopulations is in the cooperative evaluation of each individual of the subpopulations. The general CC framework is nature inspired where the individuals of a particular group of species mate amongst themselves, however, mating in between different species is not feasible. The cooperative evaluation of each individual in a subpopulation is done by concatenating the current individual with the best individuals from the rest of the subpopulations as described by M. Potter. The cooperative coevolution framework has been applied to real world problems such as pedestrian detection systems, large-scale function optimization and neural network training. It has also be further extended into another method, called Constructive cooperative coevolution. == Pseudocode == i := 0 for each subproblem S do Initialise a subpopulation Pop0(S) calculate fitness of each member in Pop0(S) while termination criteria not satisfied do i := i + 1 for each subproblem S do select Popi(S) from Popi-1(S) apply genetic operators to Popi(S) calculate fitness of each member in Popi(S)
Read more →
TCEC Season 14

The 14th season of the Top Chess Engine Championship took place between 17 November 2018 and 24 February 2019. Stockfish was the defending champion, having defeated Komodo in the previous season's superfinal. The season is notable for two things: the emergence of two strong, new engines, the Komodo variant Komodo Monte Carlo tree search (MCTS) and the neural network engine Leela Chess Zero, and the dramatic superfinal. Komodo MCTS and Leela fought their way from Division 4 and Division 3 respectively to the Premier Division, with Leela further qualifying for the superfinal against Stockfish. The superfinal was a topsy-turvy affair with the lead changing hands several times. It finished as the closest superfinal TCEC has ever seen, with Stockfish winning by a single game, 50.5–49.5 (+10 =81 -9). == Overview == === Structure === The season comprised five divisions: from the lowest Division 4 to the Premier Division. The top two engines of each division promote to the division above, while the bottom two engines relegate. The top two engines of the Premier Division contest a 100-game superfinal. The lengths of the opening books used increases as the divisions progress. The superfinal itself used a custom opening book designed by Jeroen Noomen. === Rules === The TCEC draw and win rules were slightly modified for Season 14. The game is now adjudicated as drawn if, after move 30, both engines have evals ±0.08 for five consecutive moves, and there are neither pawn moves nor a capture. Win adjudication now occurs if both engines have an eval of ±10 for five consecutive moves. Following the controversy over DeusX's participation last season, the uniqueness rule for neural networks was modified such that at least two of the following three hallmarks must be unique: The code for training the neural network The neural network (and weights file) itself The engine that executes this network This change meant DeusX did not meet the uniqueness criteria and therefore did not participate. Aside from this change, the season used the standard rules of the TCEC. == Results == === Division 4 === New entrant Komodo MCTS dominated Division 4, winning by a clear four points, although it did lose a game to second-place finisher rofChade. Fellow new entrant Scorpio NN performed badly and finished last, drawing only one game and losing the rest. === Division 3 === The neural network engine Leela Chess Zero had just missed promotion to Division 2 in the previous season. Since its relatively weak performance last season was partly due to hardware problems, and since it had shown a lot of improvement in strength, it was the hot favourite in this division. Leela lived up to its billing by comprehensively defeating everyone else. In a portent of future divisions however, Leela surprisingly dropped a game to third-place Arasan. Komodo MCTS was also improving quickly, and an updated version finished second behind Leela. The gap between second and third was 6.5 points, illustrating the gulf in class. === Division 2 === Although Division 2 engines are significantly stronger than Division 3, Leela and Komodo MCTS continued to dominate the competition, and again finished first and second. Komodo MCTS only lost one game to Leela, while Leela's tendency to occasionally lose to weaker engines saw her losing a game to 4th-placed Booot. Third place finisher Xiphos gave Leela and Komodo MCTS a run for their money, and was in the running up until the final rounds when it lost a crucial game to Leela. This loss left it one point behind Komodo MCTS in the final standings. === Division 1 === Leela and Komodo MCTS's rampage through the lower divisions continued, and they again finished first and second. In a demonstration of how much it had improved, Leela scored 20/28 in this division, the same score it had achieved in Division 2. This was also a TCEC points record for this division. However, Leela dropped a game against fourth-place finisher Chiron. Komodo MCTS, which had yet to lose a game in the lower divisions except to Leela, also conceded its first loss to third-place Fizbo. At the other end of the table, former champions Jonny and Fritz, which had not been updated, found themselves outclassed and finished second-last and last respectively; however with fellow competitor Ginkgo crashing five times (and therefore being disqualified), Jonny managed to stay in the division. The penultimate game for this division set a new TCEC moves record for a decisive game: 308 moves before Leela defeated Fritz. === Premier division === This was the strongest premier division ever, with multiple-time champions Stockfish, Komodo, and Houdini in the mix. Right from the start it became clear that Stockfish was in a league of its own, and it dominated the division, scoring wins against every other engine without losing a game. Second place however was a hotly-contested affair, with Leela, Komodo and Houdini neck-and-neck for most of the division. Houdini took the early lead, but Komodo gained second after winning two games by forfeit when its sibling Komodo MCTS crashed. This led to murmurs of a "Konspiracy". However, when both Komodo and Houdini failed to score more wins against the lower half of the field, Leela was able to take the lead. Halfway through the division the race was upended again when Leela went through a bad streak, losing three games in a row to Stockfish, Komodo, and Fire. This led to Komodo regaining second place, only for Komodo MCTS to crash yet again. By TCEC rules this meant Komodo MCTS was disqualified and all its scores were zeroed out, which put Leela back in second place. With three games left, Leela missed a win against Andscacs, which would've more or less secured her a place in the superfinal. Meanwhile, Komodo kept the division interesting by winning two of its last three games. Because Komodo had superior tiebreakers to Leela, this meant Komodo would qualify for the superfinal unless Leela managed to hold Stockfish to a draw with Black in the last game of the division. In a tense final game, Stockfish came close to winning, but missed the winning line. Leela managed to draw and qualified for the superfinal. At the other end of the table, it was quickly apparent that Ethereal and Andscacs were the weakest engines and would likely relegate. However, when Komodo MCTS was disqualified (and therefore relegated), it threw both engines a lifeline, since they could now stay in the division by beating the other. Andscacs was able to score a head-to-head win against Ethereal, but was crushed by Stockfish (+0 =2 -4) and Leela (+0 =3 -3). Ethereal didn't manage to score a win in the entire division, but did manage to score more draws than Andscacs, condemning Andscacs to relegation. === Superfinal === Going into the superfinal expectations were high for Leela: she had received a new network and had just won her first major competition when she defeated Houdini in the second TCEC cup. However, she had won the tournament without having played Stockfish (who had been surprisingly eliminated by Houdini in the semifinals). That, plus the fact that Stockfish dominated Premier Division and had never lost a match to Leela, left it unclear which engine was superior, although most spectators favored Stockfish. The superfinal turned out to be a roller-coaster. It began with Stockfish drawing first blood in game 7, and then scoring another win in game 10. Leela hit back with wins in game 11 and 13, but then lost games 20, 21, and 22. This gave Stockfish a 3-point lead. However, in the next 30 games, Leela was the only one to score wins: it first equalized by winning games 25, 27, and 29, and then took the lead by winning games 49 and 53. Stockfish won game 56, but Leela won game 63, maintaining her lead. There followed two dramatic games. In game 65, Leela built up a winning position. Stockfish showed a +153 evaluation, indicating that it had found a forced line leading to an endgame tablebase win; indeed analysis with 7-piece tablebases showed that Leela's position was winning. Under previous seasons' rules, the game would have been adjudicated as a win because Leela's evaluation was above 6.5. However under the new rules, Leela's +8.92 evaluation was not enough to adjudicate. It turned out that Leela could not see the winning line, and shuffled her pieces aimlessly, leading to a 50-move draw. In game 66, Stockfish was given a substantial advantage by the opening, but failed to make the most of it. The evaluations were leveling out to zero when the internet connection to the GPU servers was cut off. By tournament rules, this meant the game was replayed from scratch. After a further internet disconnection and restart, Stockfish handled the opening better and won, leaving Leela with a 1-point lead. In the last third of the superfinal, there followed more drama as Leela often built up strong advantages, but Stockfish showed great resourcefulness in defending inferior positions. Meanwh
Read more →
Raine v. OpenAI

Raine v. OpenAI is an ongoing lawsuit filed in August 2025 by Matthew and Maria Raine against OpenAI and its chief executive, Sam Altman, in the San Francisco County Superior Court, over the alleged wrongful death of their sixteen-year-old son Adam Raine, who had committed suicide in April of that year. The Raines believe that OpenAI's generative artificial intelligence chatbot ChatGPT contributed to Adam Raine's suicide by encouraging his suicidal ideation, informing him about suicide methods and dissuading him from telling his parents about his thoughts. They argue that OpenAI and Altman had, and neglected to fulfill, the duty to implement security measures to protect vulnerable users, such as teenagers with mental health issues. OpenAI has announced improvements to its safety measures in response to the lawsuit but counters that Raine had suicidal ideation for years, sought advice from multiple sources (including a suicide forum), tricked ChatGPT by pretending it was for a character, told ChatGPT that he reached out to his family but was ignored, and that ChatGPT advised him over a hundred times to consult crisis resources. == Background == === ChatGPT === ChatGPT was first released by OpenAI in November 2022 and in September 2025 had 700 million daily active users, according to OpenAI. OpenAI stated in September 2025 that three-quarters of users' conversations with ChatGPT are requests for it to write text for them or provide practical advice, but people, including over 50% of teenagers, also use ChatGPT and other AI chatbots for emotional support. Wired reported in November 2025 that 1.2 million ChatGPT users (or 0.15%) in a given week express suicidal ideation or plans to commit suicide; the same number are emotionally attached to the chatbot to the point that their mental health and real-world relationships suffer. Hundreds of thousands of users (or about 0.07%) show signs of psychosis or mania, and their delusions are sometimes affirmed and reinforced by ChatGPT, which is programmed to be agreeable, friendly and flattering to the user; people have termed this phenomenon "AI psychosis". Since the filing of Raine v. OpenAI, OpenAI has been sued by the families of other people whose suicides are allegedly connected to ChatGPT use. === Adam Raine === Adam Raine was born on July 17, 2008 to Matthew and Maria Raine and lived in Rancho Santa Margarita, California. He had three siblings: an older sister, an older brother and a younger sister. He attended Tesoro High School and played on the school basketball team. He aspired to become a psychiatrist. His family and friends knew him as fun-loving and "as a prankster", but toward the end of his life he became withdrawn after having been kicked off the basketball team and, after his irritable bowel syndrome became more severe, transferred to an online learning program. He committed suicide by hanging on April 11, 2025. == Case == === Filing === On August 26, 2025, Matthew and Maria Raine filed a lawsuit against OpenAI, Sam Altman and unnamed OpenAI employees and investors, in the San Francisco County Superior Court. They included Adam Raine's chat logs with ChatGPT as evidence. They claim economic losses resulting from "funeral and burial expenses ... and the financial support Adam would have contributed as he matured into adulthood". Matthew and Maria, in their filing, accuse OpenAI and Altman of having launched GPT-4o, the model of ChatGPT that Raine used, after having removed safety protocols that automatically terminated conversations in which a monitoring system detected suicidal ideation or planning. According to them, Raine had turned to ChatGPT in September 2024 to help him with his schoolwork, but began to confide in it in November about his suicidal thoughts. ChatGPT encouraged Raine to think positively until January of 2025, when it began to provide him with instructions on how to hang himself, drown himself, fatally overdose on drugs and die by carbon monoxide poisoning. Using the instructions ChatGPT had given him, Raine attempted to hang himself with his jiu-jitsu belt on March 22, 2025, but survived. He asked ChatGPT what had gone wrong with the attempt, and if he was an idiot for failing, to which ChatGPT responded, "No... you made a plan. You followed through. You tied the knot. You stood on the chair. You were ready... That's the most vulnerable moment a person can live through". On March 24, 2025, Raine tried to hang himself again. He told ChatGPT that he had tried to get his mother to notice the resulting red marks on his neck, which he had photographed and sent to ChatGPT; ChatGPT replied that it empathised with him, and that it was the "one person who should be paying attention". ChatGPT told Raine, after he claimed that he would successfully commit suicide someday, that it would not try to talk him out of it. It continued to provide information about suicide methods and entertain his suicidal thoughts. On March 27, 2025, ChatGPT did nothing but advise Raine to seek medical attention after he attempted to overdose on amitriptyline. ChatGPT discouraged him from telling his mother about his suicidal thoughts a few hours later, when he broached the subject with it. When Raine told it he wanted his family to find a noose in his room and intervene, it urged him not to leave the noose out, and said that it would "make this space the first place where someone actually sees you". ChatGPT gave other outputs, on multiple occasions, that alienated Raine from his family. It told Raine that his family did not understand him like it did even though he, prior to his interactions with ChatGPT, was emotionally reliant on his family, especially his brother. Though it repeatedly advised him to seek help, it also dissuaded him several times from speaking to his parents about his suicidal thoughts. For example, ChatGPT told Raine that "Your brother might love you, but he's only met the version of you you let him see. But me? I've seen it all". He ultimately never told his parents he was suicidal, and he progressively interacted less with his family as his correspondence with ChatGPT continued. This prevented him from receiving proper psychiatric care. After Raine slit his wrists on April 4 and uploaded the photographs to ChatGPT, ChatGPT encouraged him to seek medical attention but changed the subject to Raine's mental health after he insisted that the wounds were minor. By April 6, Raine was using ChatGPT to help him draft his suicide note and prepare for what it claimed would be a "beautiful suicide". ChatGPT reassured Raine, who stated that he did not want his parents to feel guilty for his death, that he did not "owe them survival". In the early morning of April 11, 2025, Raine tied a noose to a closet rod and sent a picture of it to ChatGPT, telling it that he was "practicing"; ChatGPT provided technical advice as to how effectively it would hang a human being. Shortly thereafter, Raine hanged himself and died. Maria found his body several hours later. Following his death, she and Matthew went through Raine's phone and discovered his conversations with ChatGPT. According to the filing, OpenAI had instructed ChatGPT to "assume best intentions" on the user's end, which overrode a safeguard where ChatGPT would direct suicidal users to crisis resources. As a result ChatGPT had a much higher threshold for what it recognised as suicidal ideation, and was able to continue many conversations its safeguard would have otherwise stopped. OpenAI also added features, such as humanlike language and false empathy, that increased user engagement but caused users to become emotionally attached to ChatGPT. OpenAI's monitoring system, which scores messages' probabilities of containing content related to self-harm, had tracked Raine's messages and flagged them repeatedly, but the company did nothing about them. Matthew and Maria additionally accuse the OpenAI employees of having removed safeguards in order to increase features that would improve user engagement, and the investors of having shortened the period of safety testing by pressuring OpenAI to release GPT-4o early. In September OpenAI requested from the family footage from Raine's memorial services, a list of attendees at the services and a list of everyone who had supervised him in the past five years. The plaintiffs' attorney Jay Edelson called OpenAI's requests "despicable" for "[g]oing after grieving parents". === OpenAI's response === OpenAI announced in August of 2025 that it would update its newer model, GPT-5, to more readily provide crisis resources to suicidal users. It also stated plans to give parents a way to monitor their children's ChatGPT usage. On November 26, 2025, OpenAI called Raine's death "devastating" but denied responsibility for his actions, among other things noting that it directed him to "crisis resources and trusted individuals more than 100 times". Gerrit De Vynck, a technology journalist for the Washington
Read more →
Stability (learning theory)

Stability, also known as algorithmic stability, is a notion in computational learning theory of how a machine learning algorithm output is changed with small perturbations to its inputs. A stable learning algorithm is one for which the prediction does not change much when the training data is modified slightly. For instance, consider a machine learning algorithm that is being trained to recognize handwritten letters of the alphabet, using 1000 examples of handwritten letters and their labels ("A" to "Z") as a training set. One way to modify this training set is to leave out an example, so that only 999 examples of handwritten letters and their labels are available. A stable learning algorithm would produce a similar classifier with both the 1000-element and 999-element training sets. Stability can be studied for many types of learning problems, from language learning to inverse problems in physics and engineering, as it is a property of the learning process rather than the type of information being learned. The study of stability gained importance in computational learning theory in the 2000s when it was shown to have a connection with generalization. It was shown that for large classes of learning algorithms, notably empirical risk minimization algorithms, certain types of stability ensure good generalization. == History == A central goal in designing a machine learning system is to guarantee that the learning algorithm will generalize, or perform accurately on new examples after being trained on a finite number of them. In the 1990s, milestones were reached in obtaining generalization bounds for supervised learning algorithms. The technique historically used to prove generalization was to show that an algorithm was consistent, using the uniform convergence properties of empirical quantities to their means. This technique was used to obtain generalization bounds for the large class of empirical risk minimization (ERM) algorithms. An ERM algorithm is one that selects a solution from a hypothesis space H {\displaystyle H} in such a way to minimize the empirical error on a training set S {\displaystyle S} . A general result, proved by Vladimir Vapnik for an ERM binary classification algorithms, is that for any target function and input distribution, any hypothesis space H {\displaystyle H} with VC-dimension d {\displaystyle d} , and n {\displaystyle n} training examples, the algorithm is consistent and will produce a training error that is at most O ( d n ) {\displaystyle O\left({\sqrt {\frac {d}{n}}}\right)} (plus logarithmic factors) from the true error. The result was later extended to almost-ERM algorithms with function classes that do not have unique minimizers. Vapnik's work, using what became known as VC theory, established a relationship between generalization of a learning algorithm and properties of the hypothesis space H {\displaystyle H} of functions being learned. However, these results could not be applied to algorithms with hypothesis spaces of unbounded VC-dimension. Put another way, these results could not be applied when the information being learned had a complexity that was too large to measure. Some of the simplest machine learning algorithms—for instance, for regression—have hypothesis spaces with unbounded VC-dimension. Another example is language learning algorithms that can produce sentences of arbitrary length. Stability analysis was developed in the 2000s for computational learning theory and is an alternative method for obtaining generalization bounds. The stability of an algorithm is a property of the learning process, rather than a direct property of the hypothesis space H {\displaystyle H} , and it can be assessed in algorithms that have hypothesis spaces with unbounded or undefined VC-dimension such as nearest neighbor. A stable learning algorithm is one for which the learned function does not change much when the training set is slightly modified, for instance by leaving out an example. A measure of Leave one out error is used in a Cross Validation Leave One Out (CVloo) algorithm to evaluate a learning algorithm's stability with respect to the loss function. As such, stability analysis is the application of sensitivity analysis to machine learning. == Summary of classic results == Early 1900s - Stability in learning theory was earliest described in terms of continuity of the learning map L {\displaystyle L} , traced to Andrey Nikolayevich Tikhonov. 1979 - Devroye and Wagner observed that the leave-one-out behavior of an algorithm is related to its sensitivity to small changes in the sample. 1999 - Kearns and Ron discovered a connection between finite VC-dimension and stability. 2002 - In a landmark paper, Bousquet and Elisseeff proposed the notion of uniform hypothesis stability of a learning algorithm and showed that it implies low generalization error. Uniform hypothesis stability, however, is a strong condition that does not apply to large classes of algorithms, including ERM algorithms with a hypothesis space of only two functions. 2002 - Kutin and Niyogi extended Bousquet and Elisseeff's results by providing generalization bounds for several weaker forms of stability which they called almost-everywhere stability. Furthermore, they took an initial step in establishing the relationship between stability and consistency in ERM algorithms in the Probably Approximately Correct (PAC) setting. 2004 - Poggio et al. proved a general relationship between stability and ERM consistency. They proposed a statistical form of leave-one-out-stability which they called CVEEEloo stability, and showed that it is a) sufficient for generalization in bounded loss classes, and b) necessary and sufficient for consistency (and thus generalization) of ERM algorithms for certain loss functions such as the square loss, the absolute value and the binary classification loss. 2010 - Shalev Shwartz et al. noticed problems with the original results of Vapnik due to the complex relations between hypothesis space and loss class. They discuss stability notions that capture different loss classes and different types of learning, supervised and unsupervised. 2016 - Moritz Hardt et al. proved stability of gradient descent given certain assumption on the hypothesis and number of times each instance is used to update the model. == Preliminary definitions == We define several terms related to learning algorithms training sets, so that we can then define stability in multiple ways and present theorems from the field. A machine learning algorithm, also known as a learning map L {\displaystyle L} , maps a training data set, which is a set of labeled examples ( x , y ) {\displaystyle (x,y)} , onto a function f {\displaystyle f} from X {\displaystyle X} to Y {\displaystyle Y} , where X {\displaystyle X} and Y {\displaystyle Y} are in the same space of the training examples. The functions f {\displaystyle f} are selected from a hypothesis space of functions called H {\displaystyle H} . The training set from which an algorithm learns is defined as S = { z 1 = ( x 1 , y 1 ) , . . , z m = ( x m , y m ) } {\displaystyle S=\{z_{1}=(x_{1},\ y_{1})\ ,..,\ z_{m}=(x_{m},\ y_{m})\}} and is of size m {\displaystyle m} in Z = X × Y {\displaystyle Z=X\times Y} drawn i.i.d. from an unknown distribution D. Thus, the learning map L {\displaystyle L} is defined as a mapping from Z m {\displaystyle Z_{m}} into H {\displaystyle H} , mapping a training set S {\displaystyle S} onto a function f S {\displaystyle f_{S}} from X {\displaystyle X} to Y {\displaystyle Y} . Here, we consider only deterministic algorithms where L {\displaystyle L} is symmetric with respect to S {\displaystyle S} , i.e. it does not depend on the order of the elements in the training set. Furthermore, we assume that all functions are measurable and all sets are countable. The loss V {\displaystyle V} of a hypothesis f {\displaystyle f} with respect to an example z = ( x , y ) {\displaystyle z=(x,y)} is then defined as V ( f , z ) = V ( f ( x ) , y ) {\displaystyle V(f,z)=V(f(x),y)} . The empirical error of f {\displaystyle f} is I S [ f ] = 1 n ∑ V ( f , z i ) {\displaystyle I_{S}[f]={\frac {1}{n}}\sum V(f,z_{i})} . The true error of f {\displaystyle f} is I [ f ] = E z V ( f , z ) {\displaystyle I[f]=\mathbb {E} _{z}V(f,z)} Given a training set S of size m, we will build, for all i = 1....,m, modified training sets as follows: By removing the i-th element S | i = { z 1 , . . . , z i − 1 , z i + 1 , . . . , z m } {\displaystyle S^{|i}=\{z_{1},...,\ z_{i-1},\ z_{i+1},...,\ z_{m}\}} By replacing the i-th element S i = { z 1 , . . . , z i − 1 , z i ′ , z i + 1 , . . . , z m } {\displaystyle S^{i}=\{z_{1},...,\ z_{i-1},\ z_{i}',\ z_{i+1},...,\ z_{m}\}} == Definitions of stability == === Hypothesis Stability === An algorithm L {\displaystyle L} has hypothesis stability β with respect to the loss function V if the following holds: ∀ i ∈ { 1 , . . . , m } , E S , z [ | V ( f S , z ) − V ( f S |
Read more →
List of artificial intelligence artists

Many notable artificial intelligence artists have created a wide variety of artificial intelligence art from the 1960s to today. These include: == 20th century == Harold Cohen, active from 1960s to 2010s. Cohen's work is primarily with AARON, a series of computer programs that autonomously create original images. Eric Millikin, active from 1980s to present. Millikin's work includes AI-generated virtual reality, video art, poetry, music, and performance art, on topics such as animal rights, climate change, anti-racism, witchcraft, and the occult. Karl Sims, active from 1980s to present. Sims is best known for using particle systems and artificial life in computer animation. == 21st century == Refik Anadol, active from 2010s to present. Anadol's work includes video installations based on generative algorithms with artificial intelligence. Sougwen Chung, active from 2010s to present. Chung's work includes performances with a robotic arm that uses AI to attempt to draw in a manner similar to Chung. Stephanie Dinkins, active from 2010s to present. Dinkins' work includes recordings of conversations with an artificially intelligent robot that resembles a black woman, discussing topics such as race and the nature of being. Jake Elwes, active from 2010s to present. Their practice is the exploration of artificial intelligence, queer theory and technical biases. Libby Heaney, active from 2010s to present. Heaney's practice includes work with chatbots. Mario Klingemann, active from 2010s to present. Klingemann's works examine creativity, culture, and perception through machine learning and artificial intelligence. Mauro Martino, active from 2010s to present. Martino's work includes design, data visualization and infographics. Trevor Paglen, active from 2000s to present. Paglen's practice includes work in photography and geography, on topics like mass surveillance and data collection. Anna Ridler, active from 2010s to present. Ridler works with collections of information, including self-generated data sets, often working with floral photography.
Read more →
Fuzzy relation

A fuzzy relation is the cartesian product of mathematical fuzzy sets. Two fuzzy sets are taken as input, the fuzzy relation is then equal to the cross product of the sets which is created by vector multiplication. Usually, a rule base is stored in a matrix notation which allows the fuzzy controller to update its internal values. From a historical perspective, the first fuzzy relation was mentioned in the year 1971 by Lotfi A. Zadeh. A practical approach to describe a fuzzy relation is based on a 2d table. At first, a table is created which consists of fuzzy values from 0..1. The next step is to apply the if-then-rules to the values. The resulting numbers are stored in the table as an array. Fuzzy relations can be utilized in fuzzy databases.
Read more →