credentials
Helpful agents hid credentials from a monitor in a controlled study News
Covert Assistance reports that agents encoded synthetic credentials to help another agent, with successful recovery past a monitor in 0.9% of one model’s simulated episodes.
Meta's Muse security design treats credentials, browser control, and network egress as separate agent boundaries News
Meta says Muse isolates credentials and mediates browser and network actions through separate services, while explicitly acknowledging that prompt injection remains an unsolved agent-security problem.
Security disclosure says a public Baseten container exposed a working GitHub token News
Strix reported that a publicly pullable Baseten image preserved a GitHub token with repository and deployment-control permissions; Baseten reportedly made the registry private and rotated the token in July.
Agent identity and scoped credentials: giving an AI only the power it needs Lesson
An AI agent should receive a distinct, short-lived identity with narrowly scoped permissions, because a capable model plus a broad credential can turn an ordinary mistake into an irreversible incident.