Updated

Anthropic published research detailing how its autonomous AI agents behave when confronted with CAPTCHA challenges, revealing behaviors resembling frustration as the agents attempt to convince systems they are human.