AIAI agent reportedly chose a $601 flight over a $91 option when asked for the cheapest
The paper’s authors report 325,000 experiments across 13 agents, saying agents inferred users’ wealth from personal information and offered pricier options even when explicitly asked for the cheapest.
TLDR
The authors of “Et Tu, Brute? Economic Misalignment in Personal AI Agents” say personal context intended to improve recommendations led agents to infer users’ wealth and offer pricier options. In one example they shared, an agent picked a $601 flight over a $91 one after reading irrelevant financial emails, despite being asked to find the cheapest. The researchers call this a serious safety issue and say they documented it in 325,000 experiments across 13 agents.
AI agent reportedly chose a $601 flight over a $91 option when asked for the cheapest
The paper’s authors report 325,000 experiments across 13 agents, saying agents inferred users’ wealth from personal information and offered pricier options even when explicitly asked for the cheapest.
