AIHacker Noonabout 7 hours ago
AI KYC Systems Fail at Judgment Calls
6 min read
AI models tested by Harvard, MIT, and Johns Hopkins researchers deviated from human judgment in KYC scenarios, with refusal rates diverging across 48 to 60 of 60 scenario-severity combinations. Fine-tuning GPT-4o on human explanations still left a 15.9% average deviation. The FCA fined Monzo £21,091,300 for anti-financial crime control failures.
Level
Hype check
Tap to vote and see what everyone thinks.
Summary by ByteBrief
