Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

What happens when a state-of-the-art AI assistant can read your email, browse the web, and fill in your passwords — but can’t reliably tell a scam from the real thing? In this video, you’ll see real examples of frontier AI agents: These aren’t edge cases. This is the result of 1Password’s new benchmark: SCAM — Security Comprehension & Awareness Measure.

When AI acts on your behalf, auditability is non-negotiable. In the latest Principles in Practice video, Anand Srinivas, 1Password VP of Product & AI, explains why every AI agent action involving credentials must leave a clear audit trail: Who approved the access and why When and where were credentials used What did the agent access and when Did access end when the task was completed Without auditability, AI usage can break trust between employees, security teams, customers, and regulators.

We worked with the Autonomy team to show how 1Password can secure agent swarms using a safer pattern: just-in-time, least-privilege access, without inheriting broad device, cloud, or infrastructure permissions, and without hardcoding secrets into agents.