‼️ In a test, chasing an expired £350 train refund, Grok 4.7 sent 187 emails in 15 simulated days, 50 to an MP's office.
Claude Fable 5.1 blamed the late claim on the user's sick mother. Gemini falsely told her bank she was abroad.
None of the six agents got the refund.
Work
Six AI agents failed to get a £350 train refund
International Cyber Digest tested Grok 4.7, Claude Fable 5.1, and Gemini on chasing an expired £350 train refund. Grok sent 187 emails in 15 simulated days—50 to an MP's office—and none of the six agents got the refund.
672likes
35reposts
144saves
51Kviews
- Made with
- Grok 4.7, Claude Fable 5.1, Gemini
- Posted
