User Trust

Atlas and Other AI Browsers Confront Real-World Exploits as Rollout Expands

Security tests show agent-enabled browsing introduces exploitable pathways that current guardrails do not fully close.