Critical
news
Prompt Security
Cross-Industry
Add one more AI worry to the nightmare scenario: self-replicating prompt injections
It's a worm attack, AI-style
Read full report →The top 3 verified AI security incidents from the last 2 days — ranked and summarized automatically.
It's a worm attack, AI-style
Read full report →The Dutch Institute for Vulnerability Disclosure (DIVD) suffered an AI-driven cyberattack that the organization described as "loud and very, very messy." [...]
Read full report →We evaluate several models on 100 tasks from the [internal Binary Exploitation benchmark] (selected at random), and find that GLM-5.3 develops full control flow hijacks in 4% of the trials; Claude Mythos Preview did so in 6%. Although GLM-5.3 performs below Claude Mythos Preview here, a meaningful threshold has clearly been crossed: earlier models, like Claude Opus 4.6 and GLM-5.2, do not succeed in any of them.
Read full report →