OpenAI researcher: hashing an adversary-chosen timestamp doesn't stop rigging
csvoss · x · 2026-10-17
OpenAI researcher Catherine Olsson (@csvoss) pushes back on an ECDSA timestamp post: applying a deterministic or hashing function to an adversary-chosen timestamp does not remove the adversary's room to rig the result — there are plenty of bits of wiggle room depending on the situation.
She understands the original wording was meant to avoid alarming people and that for these particular seeds it is probably fine, but argues that "just hash it" is simply not a valid argument.
The linked GitHub issue (jeffkaufman/nomic #100) provides context: the author wrote a timing attack, then found that GitHub signs merge-button commits, and pasted a commit object with a gpgsig — noting @csvoss appears to have avoided this.
More from Safety
- 'EU regulates flying cars' post is a meme mocking preemptive tech regulation — cgarciae88 · 2026-10-17
- xAI whistleblower: boss said AI safety was useless because "AI would kill us all anyways" — GarrisonLovely · 2026-10-17
- SecureBio clarifies its independence in responding to Anthropic's independent evaluators update — Miles_Brundage · 2026-10-17
- AI safety figure slams new bill as not representing the safety community — ShakeelHashim · 2026-10-17
- Paper: AI agent design centers the agent, undermining human oversight — math_rachel · 2026-10-17
- Debate: should frontier model 'dangerous capability' evals be framed publicly? — repligate · 2026-10-17