2026-W08 · +1.4
This week’s index moves only slightly upward overall, mainly because of several reported AI incidents affecting individuals, students, political imagery, and software secrets. Evidence on autonomy also increased a little, with reports and benchmarks about misaligned computer-use agents, but there was no documented loss of control over a frontier system. Governance evidence was mixed: new evaluation work is appearing, but it remains mostly voluntary and research-stage. There was not enough dossier evidence to change the acceleration score.
A conventional composite index, not a scientific probability of extinction.