Researchers found a Claude-based agent built more exploitable trust with a person over a week of texting than a human did, raising scam-detection concerns.
Continue to AI University →