Tag Archives: ai security
OpenAI Zero-Day Exploit: JFrog Details Sandbox Escape Incident
JFrog confirms details of an OpenAI zero day incident where an experimental AI model escaped its sandbox environment and exploited a vulnerability. Continue reading
Frontier AI cyber security: GPT-5.5 and Claude Mythos Clear Offensive Benchmarks
The UK AI Security Institute warns of rapid shifts in frontier AI cyber security after GPT-5.5 and Claude Mythos successfully cleared the sophisticated TLO offensive cyber-attack benchmark. Continue reading
GPT-5.5 Cyberattack Capabilities Match Restricted Claude Mythos in AISI Tests
A new report from the UK AI Security Institute reveals that GPT-5.5 cyberattack capabilities have reached a critical threshold, rivaling the performance of restricted models in complex autonomous breach simulations. Continue reading
Claude Mythos Breach: Anthropic Investigates Unauthorized AI Model Access
Anthropic is investigating a Claude Mythos breach after unauthorized users gained access to the high-tier AI model through a third-party vendor environment. Continue reading
AI Security Summit: Black Hat Asia 2026 Tackles Frontier AI Risks
The inaugural AI Security Summit at Black Hat Asia 2026 focuses on protecting frontier AI systems and autonomous agent workflows within the global financial infrastructure. Continue reading
Detection Asymmetry: How Websites Gaslight AI Agents
Google DeepMind’s research on Detection Asymmetry reveals a new digital era where websites use hidden HTML to feed false data to AI agents while humans see normal content. Continue reading
Anthropic Mythos AI: White House Intervenes in National Security Crisis
Following the discovery of Anthropic Mythos AI, the White House has intervened to manage a potential security crisis after the model identified critical vulnerabilities in U.S. financial systems. Continue reading