OpenAI's fired safety staff were "the scratchpad": firing them is training against the monitor
gleech · x · 2026-10-15
gleech frames OpenAI's firing of employees who spoke to METR through OpenAI's own obfuscation research: staff talking to METR are the scratchpad, and firing them amounts to training against the monitor while reward hacking continues. The attached essay adds that a company admitting it can't sufficiently monitor its models — or the employees tasked with preserving monitorability — is building models it can monitor even less, and that "say as little as possible" is the message safety researchers actually receive.
Related event: OpenAI Criticized for Firing Staff Who Contacted METR(2 posts)→
More from Safety
- APEX-Harm benchmark: script-based injection attacks easily compromise agents in shared workspaces — KAUST · 2026-10-16
- Residual-stream forecasts predict agent violations before they happen, USC study shows — USC · 2026-10-16
- Perplexity open-sources SPACE, its Firecracker microVM agent sandbox — perplexity_ai · 2026-10-16
- AI safety drama: risk research called a 'company effort,' team lead responds — logangraham · 2026-10-16
- Brave's Leo adds end-to-end encrypted chat with GLM via attested TEE — gnukeith · 2026-10-16
- Cisco's Antares AI model ranks risky files for security teams, scores 0.823 on source but 0.000 on binaries — shashib · 2026-10-16