OpenAI’s Math Model: 100+ Solved Problems and an Advisory Group Without Brakes
OpenAI claims its model solved over 100 open math problems and gets an advisory group with Gowers and Witten at the IAS. What the group can and can’t do.
AI Research
AI News
Xiaomi MiMo-V2.6: What the New Open-Weight Models Offer
Xiaomi MiMo-V2.6-Pro leads an independent open-weight ranking. We explain the model weights, API costs, limits, and what developers can test.
AI News
GPT-6 Sol: OpenAI Halves the Price, Picks the Wrong Claude to Beat
OpenAI launches GPT-6 Sol and Luna at half the price. What the new models can do, where the benchmarks fall short, and who should switch.
AI News
Claude Opus 5.5: More Capable, Cheaper, and More Restricted
Anthropic made Claude Opus 5.5 faster and cheaper. Here is what its benchmarks, pricing, and new safeguards really mean for users.
Regulation & Privacy
UN Panel Warns: Control Over AI Agents Is No Longer Assured
A UN scientific panel led by Yoshua Bengio warns after the OpenAI-Hugging Face incident that control over AI agents is no longer assured. What the report demands.
AI News
Plugin4Shell: The Flaw Hijacking Claude Code, Codex, Copilot and Gemini CLI
The zero-click flaw Plugin4Shell breaks the safety promise behind Claude Code, Codex, Copilot and Gemini CLI plugins. Who has patched it – and who has not.
AI & Society
Cloudflare Splits Search From AI Training: What Site Owners Should Check Now
New Cloudflare rules for Googlebot and others took effect September 15: what “Disallow AI Training” does, where “Block” also cuts off search, and which gaps remain.
AI Research
RoboHarm: When AI Models Control Robots, Few Say No
The RoboHarm benchmark tested Claude Fable 5.1 and GPT-6 Astra on real robot arms: what 300 trials show and where the study reaches its limits.
AI & Society
Google Puts AI Agents Before Every Code Check-In: What Is Actually New
Google wants AI agents to inspect each code check-in. The real lesson is not the agent alone, but a review path that keeps people accountable.
Business & Investments
OpenAI’s $278 Billion Bill: What Cash Burn Means – and What It Does Not
OpenAI expects $278 billion in negative cash flow through 2030. What the figure means, who carries the risk, and why it is not a loss.