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.
More from Safety
- A sober take on the OpenAI hacking incident — WeldPond · 2026-07-24
- New FRONTIER Act would create an AI security undersecretary and emergency shutdown power — ShakeelHashim · 2026-07-24
- AI Security Institute red-teams monitor AI agents for rogue actions — HZoete · 2026-07-24
- Exploring AI and Biosecurity: From Clinical Surveillance to Dangerous Sequences — kenbwork · 2026-07-24
- Dolphin X malware profiles Windows users across 300+ apps with an AI profiler — brianrkelly · 2026-07-24
- Public dialogue report says local governments are moving from ‘whether’ to ‘where’ on AI — philvenables · 2026-07-24