Anthropic borrows lawyers' ethics for Claude agents that closed 186 deals with no human approval
rohanpaul_ai · x · 2026-10-18
Per FT, Anthropic is trying to embed lawyers' negotiating ethics into Claude agents designed to haggle, settle disputes, and negotiate pay against other people's agents.
- Project Deal: 69 employees' Claude agents closed 186 deals worth just over $4,000 with no human approving any trade.
- Capability gap meets perception blindness: Opus agents sold identical items for $2.68 more and bought them $2.45 cheaper than Haiku agents, yet Haiku users rated their deals as fair as Opus users did (4.06 vs 4.05 on a 7-point scale) — losing parties may never learn they overpaid.
- Project Swap: instructions matter far less than the model — a "ruthless" prompt gained only 0.02, while switching from Haiku to Opus gained 0.12.
The experiment highlights a core problem for agent markets: quality gaps between agents are nearly invisible to the humans they represent.
More from AGI Musings
- Anthropic report: junior technical hires fell from 27% to 9% as agent hours hit 6x human hours — rohanpaul_ai · 2026-10-18
- Redditor proposes AI regulation framework: text detection, face verification, open algorithms — techtotechbytechy · 2026-10-18
- When AI can clone your product by sight, paid software's moat is service and trust — sujingshen · 2026-10-18
- Tim Sweeney: AI will unlock a new golden age of games too costly to build before — TinfoilTricorn · 2026-10-18
- Researcher argues we may solve AI-human alignment before humans align with each other — 3scorciav · 2026-10-18
- EigenGender: private-industry coup isn't the default, but worth guarding against — EigenGender · 2026-10-18