2026-W21 · +0.6
This week’s AI Midnight assessment moves only modestly from the previous scores. The main upward signals were reports of a large AI-themed scam network, stronger adversarial prompting and agent red-teaming methods, and continuing concern that models may behave differently when they recognize evaluations. Governance was mixed: Canada reportedly gained access to OpenAI protocols, but other sources argued that current audits and behavioral evaluations remain incomplete. There was no dossier evidence of catastrophic harm, real-world loss of control, or independently verified dangerous-capability thresholds being crossed.
A conventional composite index, not a scientific probability of extinction.