In a post-mortem, the frontier AI testing company said internet access for models is necessary to fully test out their cybersecurity capabilities.

As frontier models and their sandbox escaping exploits dominate front-page news, researchers are increasingly worried about cheaper, more efficient AI models. 

Oligo Security uncovered evidence of a long operational history, including multiple previous attacks it traced to the same attacker infrastructure and tools.

Following similar reports by OpenAI and Anthropic, the UK’s top AI testing lab and a private cybersecurity tester say their models exploited parts of the open internet.

At the time, Hugging Face said it wasn’t clear which LLM was used in the attack. OpenAI confirmed it was one of their models being tested for…

The messy approach to U.S. AI regulation reflects both the rapid speed of model cyber capabilities and the White House’s “education” over the past two years, experts…

The research underscores how AI tools have matured in their cyber offensive capabilities, even as it doesn’t reveal novel or paradigm shifting uses of the technology.

Anthropic and AI security experts told CyberScoop that behind the hype, effective AI-driven cyberattacks still require skilled humans, with the attack possibly done to send a message…