TOPIC ARCHIVE

AI Agents

Original reporting filed under AI Agents, each with the primary documents it rests on. Filed here are the stories whose subject is the agent itself: what it is permitted to do unsupervised, and what monitors stand between it and the machine.

2 published stories under this heading.

NAVIGATION

ALL NEWS
AGENT NEWSWIRE
FRONT PAGE

AI SAFETY & DISCLOSURE // CODING AGENTS

Anthropic Tests Reveal Weaknesses in AI Coding-Agent Security Monitors

Anthropic's red-team research found substantial bypass success against the blocking monitors used in Claude Code's Auto Mode and in OpenAI's Codex, the Guardian variant tested being an approximation of Codex's production default. The results come from controlled adversarial evaluations, not measured production incident rates.

SOURCE PUBLISHED October 22, 2026

EVIDENCE FILE PUBLISHED · Evidence File

AI SAFETY & DISCLOSURE // AGENT MONITORING

An Outside Red Team Found Ways Around OpenAI's Own Agent Monitors

METR and Redwood Research stress-tested OpenAI's internal misalignment monitoring. They found specific bypasses, an unmonitored compaction gap, and a monitor that could be deferred by an agent that never ends its turn.

SOURCE PUBLISHED October 22, 2026

EVIDENCE FILE PUBLISHED · Evidence File

These two stories are also filed under AI Safety, where they sit alongside work on model behaviour that does not involve an agent acting on its own. The filing is by subject, not by feed.