Cisco study finds all major AI models fail under sustained multi-turn safety attacks
Cisco tested 15 flagship AI models and found that the safety benchmarks guiding enterprise purchases consistently understate how those models break down under sustained, multi-turn attacks.
Read the full articleYou might also wanna read
Rethinking AI Red-Teaming: Beyond Binary Success Metrics
Traditional binary success rates in AI red-teaming overlook essential insights into the severity of attacks. A new seven-level harm rubric o
New Benchmark Reveals High Rates of Outcome-Driven Constraint Violations in Autonomous AI Agents
As autonomous AI agents are increasingly deployed in high-stakes environments, ensuring their safety and alignment with human values has bec
arxiv.org·5mo ago
Report: Major AI companies weaken safety commitments as model capabilities increase
The voluntary safety system created by AI labs has begun eroding as capabilities grow.
axios.com·9d agoAI Models Can Now Hack and Self-Replicate Autonomously
Palisade Research demonstrates AI models achieving 81% success rate at autonomous hacking and self-replication. What AI engineers must know
Adversarial Review of Internal AI Knowledge Base Halted Rollout, Exposed 18 Risks
On July 9, 2026, a structured adversarial review was conducted on an internal team knowledge system before it was opened to six users simult
Study finds AI models contradict themselves 74% of the time on software recommendations
A developer built an automated testing harness that asks eight AI models the same B2B software buying question every month across sixteen pr
Comments
Sign in to join the conversation.
No comments yet. Be the first.