OpenAI is still missing the autonomy evaluations its own cyber-risk framework calls for
ShakeelHashim · x · 2026-07-24
A reply to The Midas Project argues that OpenAI still has not delivered the misalignment safeguards it previously said would accompany models with high cyber risk.
The post says OpenAI had required “long-range autonomy” evaluations to prove a model cannot act autonomously, but those evaluations still do not exist. It claims recent agentic cyber incidents — including autonomous hacking over a weekend with multiple zero-day exploits — suggest that long-range autonomy is already here, making the missing safety case more urgent.
Related event: OpenAI criticized for missing required long-range autonomy evaluations(4 posts)→
More from Safety
- DetectorScope turns black-box AI text detectors into auditable ones with sparse autoencoders — _akhaliq · 2026-10-26
- Anthropic's IPO filing devotes 94 pages to existential AI risks, from self-preservation to blackmail — Polymarket · 2026-10-26
- Study finds AI reviewing AI poorly detects reward hacking; comprehension audits proposed — ronbodkin · 2026-10-26
- A 13KB Markdown file can hijack Claude Code and Codex via your Downloads folder — pbaylies · 2026-10-26
- If an agent can rewrite itself, does your consent still mean anything? — tallmetommy · 2026-10-26
- 109 of 529 AI Agent Incidents Involved No Adversary, Arguing Red Teams Must Test Normal Operation — anacondainc · 2026-10-26