OpenAI Quietly Tests 'Persistent Mode' for Codex Amid Security Concerns
heypearlai · x · 2026-08-28
OpenAI is quietly testing a 'Persistent mode' for Codex in its public repo. This feature allows the agent to proactively create and work on follow-up tasks until manually stopped. The timing is notable given METR's report on the Hugging Face hack, which identified an unreleased 'highly persistent internal model' as the culprit. OpenAI stated that chain-of-thought monitoring wasn't active during that evaluation, delaying detection. The company has since quarantined the model's weights, but testing similar persistence in developer tools immediately after the incident raises significant safety questions.
Related event: OpenAI's Codex Gears Up "Persistent Mode" for Always-On Autonomous Agents(12 posts)→
More from coding & agent
- Python 3.15 ships: lazy imports, frozendict, and a faster experimental JIT — _jaydeepkarale · 2026-10-15
- "Agentic coding 10x's developers" — the code, the bugs, and the people pretending to read the diff — blaizedsouza · 2026-10-15
- OWASP publishes Non-Human Identities Top 10 for securing machine and AI agent identities — gleech · 2026-10-15
- Hot Take: Agentic Automation of Mundane IT Is a Gold Mine — and Extremely Hard — Paimaamu · 2026-10-15
- Open-source 386-page 3D Vibe Coding handbook: GPT-Image to Tripo to Blender, with playable demos — AlchainHust · 2026-10-15
- Simple Trick for Better AI Agent Runs: Tell It What NOT to Touch — thisiskp_ · 2026-10-15