Desk
Broad artificial intelligence coverage.
OpenAI’s Navier–Stokes announcement landed inside a credit fight: NYU’s Tristan Buckmaster says the lab raced after learning of his and Anthropic’s Levent Alpöge’s related Euler work.
Google Cloud’s Accenture Gemini Enterprise Business Group buys consulting feet on the ground—up to 1,000 trained FDEs—because model share alone is not converting hyperscale CapEx into sticky enterprise ROI.
Chrome 153 lands on a two-week cadence—halving the old four-week train—as Google treats AI-boosted vuln discovery and AI-boosted attacks as reasons to shrink the N-day gap.
DeepMind turns AlphaGenome into a precomputed catalog of roughly nine billion single-letter DNA variants, aiming to make mutation triage a lookup rather than a bespoke model run.
OpenAI CFO Sarah Friar argues that GPT-6 Astra, a billion-user consumer base, and in-house inference silicon form a compounding loop of capability, demand, and cheaper serving.
NVIDIA opens two official Rust paths to native PTX kernels—SIMT via cuda-oxide and Tile via cutile-rs—betting the AI systems stack will keep migrating to Rust.
Former DeepMind researcher Danijar Hafner’s stealth robotics startup bets model-based reinforcement learning can carry game-trained planners into unfamiliar physical spaces.
OpenAI says an internal multi-agent system—stronger than GPT-6 Astra—produced a Lean-checked singularity proof for Navier–Stokes statements C and D, and it will not claim the Millennium Prize.
OpenAI Group PBC is putting up to $5 million—grants to $1 million each—behind independent studies of generative AI’s effects on teens 13–17, with applications due October 6, 2026.
Reducto collapses multi-stage agentic OCR into one full-page pass, cuts its own error rate 20%, and drops price to a flat cent a page—while keeping the weights closed and the evals internal.
OpenAI pushes ChatGPT Edu into two top j-schools and restates a years-long newsroom partnership map—access and training upstream, enterprise credits and playbooks downstream.
OpenBMB's 2.52B dense MiniCPM5-2B posts a 53.9 average across 34 benchmarks and ships Apache-2.0 weights plus training data—strongest where agents call tools, weaker on textbook knowledge.