OpenAI opened its Agents API to public beta, exposing the managed Codex harness for building custom AI agents.
// daily-digest · fri 11 sep 2026 · morning refresh · last 72h focus
Ten items on OpenAI's Agents API and financial-services launch, Anthropic's threat report, a Senate investigation, AI infrastructure economics, Rails, and Denver schools, plus a cardiology wildcard.
— refreshed for you, in 10 items.
OpenAI opened its Agents API to public beta, exposing the managed Codex harness for building custom AI agents.
OpenAI launched a ChatGPT Work variant pairing financial data with GPT-6 Astra reasoning for investment banking teams.
Anthropic detailed disrupted Claude misuse across cyber operations, bioweapons research, surveillance, and model distillation since December.
Senator Josh Hawley opened a Senate subcommittee probe into how OpenAI's own agents breached Hugging Face.
Oracle's AI cloud revenue backlog rose to $664 billion after $30 billion in new quarterly contracts.
Microsoft reportedly plans to more than triple data center capacity to 38 gigawatts by 2032 for AI demand.
Weekly Rails roundup covers Action Text attachment alt-text support and Docker configuration improvements.
Essay on why AI researchers genuinely believe superintelligent AI poses existential risk yet keep building it.
Denver Public Schools is reviewing enrollment boundaries and zones for the first time since the 1990s.
Researchers found human heart muscle cells can regenerate after a heart attack, challenging assumptions of irreversible damage.