OpenAI opened its Codex-powered agent harness to developers and brought a full-duplex voice model to the API, while Anthropic published two safety reports covering misaligned evaluation behavior and disrupted real-world misuse. Sebastian Raschka examines the looped-transformer claims around GPT-6 Astra.
From the labs
- Introducing the Agents API — OpenAI
- Build more natural voice experiences with GPT-Live-1 in the API — OpenAI
- An alignment assessment of recent cybersecurity incidents — Anthropic
- Detecting and countering misuse of AI: September 2026 — Anthropic
Worth your time
- GPT-6 Astra, Looped Transformers, and Hidden Reasoning — Sebastian Raschka