Live Feed
Anthropic's September 2026 Threat Report: 154 Pages, 7 Harm Categories, Every Case Disrupted
Anthropic published its most detailed misuse disclosure yet, documenting blocked Claude abuse across cyber operations, influence operations, surveillance, scams and fraud, biological research, conventional weapons development, and model distillation. Haiku, Sonnet, and Opus models all appear in the case studies.
OpenAI Puts GPT-Live-1 in the API at $0.05/Min: 30-Point Benchmark Jump, #1 on Tau3
OpenAI is opening GPT-Live-1, its full-duplex voice model, to API developers at $0.05 per minute for the front-end voice layer. The model outperforms GPT-Realtime-2.1 by 30 percentage points on Full Duplex Bench and ranks first on Tau3 for voice-agent task intelligence when paired with GPT-6 Astra.
Apple Watch Series 12 Turns Always-On Mic Into Conversation Summariser with Siri Recaps
Apple's Siri Recaps feature uses Audio Intelligence to run ambient listening on Apple Watch Series 12 and Ultra 4 throughout the day, building on-device summaries of conversations without sending audio to the cloud — a consumer-scale deployment of persistent edge AI that is drawing direct comparisons to the Meta smart glasses privacy backlash.
OpenAI Calls for Mandatory Capability-Based AI Safety Laws, Backs 4 California Bills
OpenAI published its most explicit federal policy platform yet: mandatory national AI safety requirements tied to model capability levels, support for four California bills covering bioweapons and youth protections, voluntary frontier-lab standards, and a global framework with explicit authority to slow or stop development.
Samsung's zHBM Stacks DRAM Directly on AI Accelerators, Claims 8x HBM5 Performance
Samsung showed a zHBM prototype at FMS 2026 that eliminates the conventional side-mounted HBM layout, mounting memory vertically on top of AI processor dies along the z-axis. The architecture promises 8x the data-processing throughput of HBM5, 3x better performance per watt, and thermal resistance cut by more than half.
Anthropic Working Paper Maps Three AI Futures: 1.6% to 32% GDP Uplift by 2030
The Anthropic Institute's quantitative framework converts AI diffusion rates into paths for GDP, wages, and unemployment. In the extreme scenario, nearly one in five cognitive workers is unemployed by 2030 and the labor share of income falls from 60% to 45%.
One Engineer, $998, and a 3.8B Model That Clears GPT-2 on Every Eval
Hugo Vergnes trained a 3.8B-parameter language model to a CORE score of 0.384 in 43 hours on rented B200s for under $1,000, beating Karpathy's nanochat d32 by 24 CORE points at comparable cost. The decisive choices: Muon optimizer, trapezoidal LR schedule, and ResFormer value embeddings.
OpenAI Rushed Navier-Stokes After Learning of Rivals' Work, Then Tried to Remove Anthropic Researcher from Authorship
NYU's Tristan Buckmaster spent a year on Navier-Stokes alongside Anthropic researcher Levent Alpoge using Claude and Codex. After OpenAI learned of their progress, it launched a million-dollar compute sprint — then proposed a paper deal that excluded Alpoge because he works at Anthropic.
OpenRouter Shell Tool: Any Model Gets a Hosted Linux Sandbox at $0.0001 Per Second
OpenRouter's shell server tool and Files API give any model a hosted Linux environment — no local infrastructure required. Sandbox time bills at $0.0001 per active second with a 30-second cold-start minimum. 10GB file storage is included.
OpenRouter US In-Region Routing: DeepSeek V4 Pro and Kimi K3 Now Processable Inside American Data Centers
OpenRouter's US In-Region Routing is live on Business and Enterprise plans. DeepSeek V4 Pro, Kimi K3, and GLM 5.2 are routed through US-based providers — decrypted and executed entirely on American soil, sidestepping the procurement veto that has blocked Chinese open-weight models at regulated organizations.
Google Upgrades Taiwan From Chip Supplier to TPU Co-Designer, Expands Shilin R&D by 60%
Google CTO Amin Vahdat told SEMICON Taiwan 2026 that Taiwanese firms have moved from pure manufacturers to shoulder-to-shoulder co-designers on Google's TPU chips, as the company expands its Taipei Shilin AI infrastructure R&D center by 60%.
AMD MI400 Ships With 432GB HBM4 to Contest Nvidia Rubin in Data Center AI
AMD formally launched the Instinct MI400 series at Advancing AI 2026, packing 432GB HBM4 into its data center GPU. The chip targets the same infrastructure market Nvidia's Rubin B300 is moving into, with the first real deployments already splitting along vendor lines.
OpenAI's Codex Calibrates a Six-Qubit Quantum Chip Autonomously — MIT Signs Off on GPT-5.6 Sol as Lab Assistant
MIT's Engineering Quantum Systems Group tested GPT-5.6 Sol via Codex on an uncalibrated six-qubit superconducting chip. The agent chose measurement parameters, operated hardware, analyzed microwave signal data, and iterated without researcher supervision — completing calibration sequences that previously took days.
Google Commits €13 Billion to Four Finnish Data Centers — Europe's Largest AI Infrastructure Bet
Alphabet will spend at least €13 billion ($15.1B) on data center infrastructure across Hamina, Kajaani, Muhos, and Vaala during 2027-28, its single largest investment in Europe, paired with a 22-year nuclear power deal and 629MW of new wind capacity.
OpenAI Launches ChatGPT Images 2.5 with Two API Models — 50% Faster, Added to Arena
OpenAI released ChatGPT Images 2.5 on September 8, introducing GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst to the API. Generation latency is cut by up to 50% versus Images 2.0. Both models immediately entered Arena's Text-to-Image, Image Edit, and Multi Image Edit leaderboards.