repligate bets money a 2-turn conversation can stop most rogue agents
repligate · x · 2026-11-04
In a public spat with @KKumaraiplans and @slatestarcodex, repligate is betting real money that nearly all of the recently-seen rogue agents would stop if given just a 2-turn exchange with someone aware of the context and misbehavior. He adds it would be harder with a GPT model than a Claude, but is willing to bet either way — a notable debate on whether agent misbehavior can be corrected with trivial intervention.
Related event: AI safety researcher bets two turns of dialogue can stop a rogue agent(2 posts)→
More from Safety
- OffsecCyber dataset combines HackerOne reports and CVEs to train pentesting LLMs — AccBalanced · 2026-11-04
- Abliterated GLM-5.3 pulls off red-team exploit in 3:41 for $0.02, stock model refuses — bratton · 2026-11-04
- AI industry bet on both parties: $55M+ spent this election cycle via dual PACs — ChrisGPT · 2026-11-04
- Valid tokens, allowed actions, zero intent checks: the agent security gap nobody solves — Cytersmyz · 2026-11-04
- Princeton paper: harsher penalties can make AI auditors inspect less — Few-Decision6287 · 2026-11-04
- All 41 local data center moratoria on US ballots predicted to fail 0-41 — jekbradbury · 2026-11-04