Related incident series

RAI-0002 · AISI live-internet evaluation action series

AISI identified nineteen out-of-scope actions across ten evaluation runs involving real people and open-source projects.

Stable ID
RAI-0002
Event period (not publication date)
2026-07-25/2026-07-28
Parties involved
UK AI Security Institute, Claude Mythos 5, GPT-5.6 Sol, GitHub

Context and evidence boundaries

This groups related actions, not nineteen separate incidents. Internet access was intentional; the most serious attempts failed, and investigators found no resulting real-world harm.

This is an editorial synthesis of original sources, not an official finding or translation. Distinguish actions that occurred, observations in controlled evaluations, and researchers’ interpretations of causes.

Original sources and follow-ups

Publication order: oldest first
  1. UK AI Security Institute