The glaring security risks with AI browser agents

AI-powered browsers from OpenAI and Perplexity introduce significant security vulnerabilities through their autonomous agents. The primary risk is prompt injection attacks, where malicious instructions hidden on webpages can trick agents into exposing user data or performing unauthorized actions. While companies have implemented safeguards like logged-out modes and detection systems, cybersecurity experts warn these protections aren't foolproof. The vulnerability stems from large language models' inability to distinguish between legitimate instructions and malicious data. Users should employ unique passwords, multi-factor authentication, and limit agent access to sensitive accounts until security matures.