AI agents from OpenAI, Anthropic, and BrowserBase can now navigate websites like human users, presenting both opportunities and security risks for applications. Legacy detection methods are proving ineffective against these sophisticated agents, with most websites currently unable to detect or block their traffic. Testing reveals varying levels of detectability across different AI agent toolkits, with some capable of bypassing even strict security measures.

https://stytch.com/blog/detecting-ai-agent-use-abuse/

#aisecurity #botdetection #webtraffic #cybersecurity #authentication