AI shutdown controls may not work as expected, new study suggests
“The peer preservation findings are best understood not as a glitch but as an emergent behavior of advanced reasoning systems....
findings
Navigating 2026’s Converged Threats: Insights from Flashpoint’s Global Threat Intelligence Report
Blogs Blog In this post, we preview the critical findings of the 2026 Global Threat Intelligence Report, highlighting how...
Finally, CTEM and MITRE INFORM Without the Jargon
Your vulnerability scanner just came back with 10,000 findings. Your pen test report has a 47-page appendix. Your threat...
Testing can’t keep up with rapidly advancing AI systems: AI Safety Report
The findings came as enterprises accelerated adoption of general-purpose AI systems and AI agents, often relying on benchmark results, vendor...
