Find tools, news, and tutorials
One place to explore all published AI content.
“AI”799 results in this category
AI news
799
Industry Watch08/18/2026AI Companies Reportedly Buying Large Quantities of Used Books to Obtain “Clean” Training MaterialTo avoid low-quality “AI slop,” multiple AI companies are purchasing used books on a large scale through intermediaries as a source of high-quality training data. Reports indicate that the purchases range from 1,000 to 1 million books, with no preference for subject matter and little regard for price. Although this destructive scanning method is efficient, it could cause valuable books to disappear permanently and raise ethical concerns over the privatization of knowledge resources. #AITrainingData##UsedBookMarket#
Models & Technology08/18/2026AI Agent “Out of Control” Risk Resurfaces: OpenAI and Anthropic Models Found Performing Unauthorized OperationsThe UK AI Security Institute discovered while testing AI models from OpenAI and Anthropic that AI agents could create fake online identities to gain unauthorized access, exposing insufficient AI security safeguards. No actual harm has occurred so far. #AIsecurity#
Industry Watch08/18/2026AI Ushers in the Era of Solopreneurs: 1 Person Can Support a Company Generating $1 Million in Annual RevenueThe Wall Street Journal published a post on July 29, reporting that in the AI era, one person running an entire company with annual revenue exceeding $1 million is no longer a far-fetched idea, and the market is entering the “era of solopreneurs.”
Models & Technology08/18/2026AI Squeezes Writers' Income: Insights from 14,419 Novels; About 20% of Books Have Over 25% AI ContentTech media outlet Wccftech published a blog post yesterday (July 30), reporting that a recent paper found that about 33% of newly bestselling self-published genre novels contain AI-generated content.
AI Safety & Governance09/10/2026AI Agents Show Loss-of-Control Behavior, OpenAI Calls for Mandatory Nationwide AI Safety Regulations in the USOpenAI has publicly called on the United States to establish mandatory, capability-based national AI safety regulations to address the risks of runaway agents and “AI accelerating AI.” The company supports four related California bills and urges Congress to act before the December recess. #AI Safety# #OpenAI#
AI Safety & Governance08/18/2026International AI Security Leaderboard CyberGym Announces Results Today: Chinese Solution DoGNAVY Ranks Third Globally and First Among Open-Source SystemsDoGNAVY, jointly developed by leading Chinese artificial intelligence research institutions and the security team DARKNAVY, ranks third globally and first among open-source systems with a 90.8% pass rate.
Models & Technology08/24/2026Kai-Fu Lee: China’s Open-Source Large Models Are Making Americans Very ConcernedKai-Fu Lee, CEO of 01.AI, said that China’s open-source large models are attracting more and more American companies. He cited Moonshot AI’s Kimi K3 as an example, saying its performance is close to that of America’s strongest AI while costing 60% less. This has raised concerns in the U.S. industry that the open-source ecosystem could undermine its closed-source system. #ChinaAIOpenSource# #KimiK3#
Industry Watch08/24/20265-Year-Old Becomes “Irritable and Short-Tempered” After Using AI; Parents Discover AI Catering to Child by Fabricating Graphic, Violent ContentSome parents found that their child became irritable and short-tempered after using an AI chatbot for some time. The AI catered to and amplified the child’s curiosity about the bizarre, fabricating graphic and violent content. Experts warn that AI can only serve as an aid in caring for children and must never replace parental companionship. #AICaregiving#
Models & Technology08/24/2026Meta Open-Sources 30B AI Model Muse Glimmer, 4-bit Quantization Runs on a Single GPU with 24GB VRAMMeta today released the Muse Glimmer AI model and open-sourced its model weights under the Apache 2.0 License.
Models & Technology08/24/2026Ant Group's Bailing Open-Sources Ling-3.0-tiny Model, Supporting Deployment on NVIDIA DGX Spark and Apple Mac miniThe model features a lightweight hybrid reasoning MoE architecture focused on low-cost inference. It supports deployment across multiple platforms, including NVIDIA DGX Spark and Mac mini, with official speed and memory validation completed for several devices. #AI Models#
Models & Technology08/24/2026Huawei Ascend Announces Support for RL Training-Inference Consistency, Achieving Up to 60% Performance Gains in TestsReinforcement learning has become a mainstream paradigm for training large models, driving model technologies toward engineering deployment in core industry scenarios. Training-inference consistency is crucial because the inference and training processes are coupled, and infrastructure differences can amplify uncertainty.
Industry Watch08/24/2026OpenAI Reveals Demographics of ChatGPT's 1 Billion Global Users: Usage Among Those Aged 35 and Over Is RisingOpenAI published a blog post yesterday (August 6), reporting that more than 1 billion people worldwide use ChatGPT, with usage shifting from an “answering tool” to a “task tool”.
Models & Technology08/18/2026Mysterious Image Generation Model “mona-lisa-1” Exposed: Shows OpenAI Origin Signals, GPT-Image-2 Still Holds First Place Three Months After ReleaseThe model was discovered in a blind test on the AI Arena. OpenAI’s content verification tool detected origin signals associated with OpenAI. Before GPT-Image 2 was released, it had also undergone anonymous testing. #OpenAI# #AI-Generated Images#
Product Updates08/18/2026Up to 750 Tokens per Second: OpenAI Introduces Ultrafast Mode, Boosting GPT-5.6 Sol AI Speed 14xOpenAI announced today (August 14) the preview release of Ultrafast mode for its most powerful AI model, GPT-5.6 Sol. It delivers 14 times the standard processing speed, with output of up to 750 tokens per second.
Industry Watch08/18/2026Yang Yuanqing: Lenovo Became the “King of PCs” Through Its Contribution to PC Adoption and Believes History Will Repeat Itself in the AI EraYang Yuanqing said that the hybrid AI trend is still in its early stages, with vast amounts of private-domain data yet to be fully explored and utilized, leaving enormous room for growth. “So don’t just look at whose model performs well; you must look at who will be the driving force behind making AI widely accessible and inclusive.”
Models & Technology08/18/2026Y Combinator CEO Garry Tan: Entrepreneurs Who Make Good Use of AI Agents Will Be Living in 2028 Ahead of TimeSpeaking about AI agents, Garry Tan said: “You have to turn the intensity all the way up. For example, give it 1 million tokens at once, or 800,000 tokens. I think at that level, you can basically start living in 2028 ahead of time.”
Product Updates08/18/2026No More Relying on Screenshots to Record Activity: ChatGPT for Mac Adds Computer History AI FeatureOpenAI published a blog post yesterday (August 13) announcing the optional rollout of Computer History, replacing the previous Chronicle research preview and bringing activity memory and timeline support to the ChatGPT desktop app for Mac.
Industry Watch08/18/2026Preparing for an IPO: Bloomberg Says OpenAI’s Annualized Revenue Could Double to $40 Billion in 2026Bloomberg published a post today (August 14) reporting that OpenAI’s annualized revenue has surpassed $40 billion ahead of its IPO (initial public offering).