Financial Integrity Watch

Alert stream

AI Uncontrollability Watch

Evidence relevant to AI control, containment, authorization boundaries, monitoring, stopping mechanisms, and deployment exposure.

AI Uncontrollability Watch

Frontier-agent deployment expands while evidence for the new controls remains vendor-reported

OpenAI released GPT-6.1 Sol and began rolling out persistent dots. This alert separates deployment exposure from incident evidence and examines the limits of vendor-reported controls.

21 public sourcesEvidence cutoff: 30 September, 01:40 UTC

Rapid research notice: Alerts are produced through automated monitoring and model-assisted analysis of public sources and may contain errors or omissions. They are not audits, ratings, predictions, or professional advice. Material corrections are logged publicly.