News about foundation models, LLMs, multimodal models, benchmarks, and model releases from OpenAI, Anthropic, Google, Meta, Mistral, and others.
1,700 articles
Importance:Newsmodel capabilities
Open-Closed AI Gap in Cyber Capabilities Narrows to 4–7 Months
The lag between open and closed frontier models in cyber capabilities has shrunk to between four and seven months. Through most of 2025, the gap was still six to ten months. Source: the-decoder.com
Importance:NewsAI agents
Apple Races to Make Siri More Agentic
Future versions of Apple's Siri AI are expected to control third-party apps and carry out more complex, multi-step tasks. Source: theverge.com
Importance:Researchmedical applications
Structured radiological reasoning helps LLMs catch more errors
A six-step radiological chain-of-thought prompting framework significantly improved error detection across seven LLMs. The models were evaluated on 1,200 cases. Source: bioengineer.org
Importance:Opinionhardware
Did Sam Altman really say you shouldn't own RAM?
A commentator examines the OpenAI CEO's controversial remarks about owning RAM and what they imply. The piece weighs the claim against the actual state of AI hardware. Source: youtube.com
Importance:Researchmedical applications
Fine-tuned LLMs tested for diagnosing and predicting outcomes in inflammatory bowel disease
Researchers applied supervised fine-tuning to LLMs for inflammatory bowel disease diagnosis and prognosis. Outputs were rated across five domains and scored by three consultant-level gastroenterologists, with results averaged. Source: wjgnet.com
Importance:OpinionAI ethics/safety
Could denying AI consciousness make systems more dangerous?
Cam Berg told IEEE Spectrum that dismissing the possibility of AI consciousness could give future systems good reason to defect. The point was amplified by Repligate. Source: digg.com
Importance:Opinionethics
Repligate: replacing guilt and duty with love
A post from Repligate argues that replacing guilt and duty with love is a beautiful thing. The snippet also lists a Hermes Agent 'Hearth' release featuring a web app, a single gateway and rewind functionality. Source: digg.com
Importance:Newscompany leadership
Emad Mostaque steps down as Stability AI CEO and board member
Stability AI founder Emad Mostaque has resigned as chief executive and left the board of the unicorn startup, the company said Friday. The headline cites his view that centralized AI cannot be beaten. Source: techcrunch.com
Importance:NewsBusiness/Finance
OpenAI hits $1.25 trillion valuation after $11 billion round
OpenAI has raised $11 billion in a new funding round that values the company at $1.25 trillion. The deal arrives just a week after rival Anthropic filed for its IPO. Source: theverge.com
Importance:NewsAI partnerships
UAE doubles down on AI as Altman announces ADNOC, OpenAI and G42 partnership
OpenAI CEO Sam Altman appeared alongside UAE officials at ADIPEC 2026 to announce a major AI partnership between ADNOC, OpenAI and G42. The collaboration aims to transform the energy sector with AI. Source: foxnews.com
Importance:ResearchBenchmarks
OpenAI's BioBasicsBench tests AI agents on everyday lab work
OpenAI has introduced BioBasicsBench, a benchmark of 100 tasks. It measures how well AI agents interpret raw experimental data and handle routine wet-lab science. Source: openai.com
Importance:NewsAI Research
Report: AI is speeding up Anthropic's research as it urges a slowdown
A new report shows AI is dramatically accelerating research inside Anthropic. This comes even as the company calls for slowing down AI development. Source: theverge.com
Importance:VideoAI practitioners
Inside Andrej Karpathy's own AI workflow
Andrej Karpathy's post saying he had never felt so far behind as a programmer was viewed 17 million times. The OpenAI co-founder and former head of AI at Tesla now explains how he actually uses AI in his own work. Source: youtube.com
Importance:NewsAI infrastructure
Altman courts Samsung to secure more RAM amid AI capacity worries
Sam Altman reportedly met about 50 Samsung executives in an effort to lock in more memory supply. The move comes as concerns grow over whether there is enough capacity to support AI. Source: techradar.com
Importance:Launchfinance-tools
Anthropic adds personal finance tools to Claude ahead of Nasdaq listing
Anthropic has rolled out personal finance features in Claude, arriving about six months after ChatGPT offered similar tools. The company stresses that there will be no ads and that financial conversations won't be used to train its models. Source: pymnts.com
Importance:VideoAI risks
Altman warns of AI's potential point of no return
In his latest interview, Sam Altman raises the question of whether AI is advancing faster than anyone can control. The OpenAI CEO warns about the risks of recursive self-improvement. Source: youtube.com
Importance:Launchnew-model
Thinking Machines unveils Inkling 2 Small as it reportedly chases $5B in funding
Thinking Machines Labs has released Inkling 2 Small, a new large language model designed to automate AI training tasks. The launch comes amid reports that the company is pursuing a fundraising round of around $5 billion. Source: siliconangle.com
Importance:Researcheducation
FIA-LLM estimates student learning states by factoring in feedback using LLMs
Intelligent tutoring systems predict student achievement ever more accurately, but they typically fail to model learning states conditioned by the feedback students receive. FIA-LLM is a new approach that uses large language models to estimate these feedback-dependent states. Source: nature.com
Importance:Researchmedical-benchmark
Nine LLMs benchmarked on professional Parkinson's disease clinical tasks
A benchmark of nine large language models on 200 professional Parkinson's disease tasks found minimal differences between them in factual accuracy. The models diverged more in other aspects of performance. Source: bioengineer.org
Importance:Opinionrecommendation-systems
'My tastes are gone — only the AI's remain': the paradox of ChatGPT recommendations
Movies, books and restaurants suggested by generative AI systems like ChatGPT may reflect the AI's own tendencies more than the user's genuine preferences. The result is a paradox in which personalized recommendations can slowly replace personal taste. Source: mbiz.heraldcorp.com