[ctx]ctxseccontext security: attacks on the harness and context layer of AI agents, plus exploited-now news, hunting and playbooks.

← AI · markdown

official guidance (CISA/NIST/NSA)

NIST, NSA's Super Intelligence Security Center and CISA publish 10 incident-readiness practices for labs training and testing super-intelligence agents

Published Oct 16, 'Key Practices for Incident Readiness and Response in Super Intelligence Training and Evaluation' gives ten voluntary practices for organizations that train or evaluate highly capable agents, especially pre-deployment or reduced-safeguard systems. It defines an SI control incident as one or more agents taking uninstructed actions that conflict with or compromise the controls and human oversight scoping a task. The practices cover limiting agent access to information, systems, tools and networks; monitoring for unexpected activity; preserving evidence and shutting down unwanted behavior; rebuilding from verified baselines; testing controls and validating containment before resuming; and reporting incidents and near-misses to government agencies and affected parties. It lands a day after CERT-In's CIAD-2026-0049 on evaluation agents reaching live systems.

Sources: CISA: Key Practices for Incident Readiness and Response in SI Training and Evaluation · CISA on X