Anthropic’s Fable 5 Model Jailbroken Within Days

Multiple Schneier blog posts highlight emergent risks from rapid AI deployment and adversary adaptation. Key items: Anthropic’s Fable 5 (Mythos) was jailbroken within days and subsequently subject to U.S. export-control restrictions that cut access broadly; malware authors are embedding policy-triggering/forbidden text in payload headers to poison or derail AI-based analysis pipelines (targeting bioinformatics and developer tooling); the U.S. federal government disclosed thousands of AI use cases raising governance and safety concerns; law-enforcement misuse of Flock surveillance cameras and a