The Last Safe Hour newshttps://lastsafehour.com/news/Research briefings on AI risk.en-usTests expose blind spots in AI monitors, and ways to improve themhttps://lastsafehour.com/news/2026-10-26-monitoring-tests/https://lastsafehour.com/news/2026-10-26-monitoring-tests/Mon, 26 Oct 2026 14:01:38 +0000Four recent studies examine monitoring blind spots, tested fixes and ways to reduce cheating during training.The week in AI risk October 19 to 23https://lastsafehour.com/news/2026-10-25-week-in-ai-risk/https://lastsafehour.com/news/2026-10-25-week-in-ai-risk/Sun, 25 Oct 2026 16:26:06 +0000Monitoring gaps, AI-assisted attacks and what four new reports tell us.