…Across the period we studied, attackers’ use of AI shifted from techniques to gain initial access to a system towards activity carried out once they were inside the system. For example, the…
Frontier Red Team Assessing Claude Mythos Preview’s cybersecurity capabilities Apr 7, 2026 Nicholas Carlini, Newton Cheng, Keane Lucas, Michael Moore, Milad Nasr, Vinay Prabhushankar, Winnie Xiao Hakeem Angulu, Evyatar Ben Asher…
…For more on our safety initiatives, see our Responsible Scaling Policy , Frontier Red Team , and Safeguards work. Footnotes Related content Agentic coding and persistent returns to expertise Paving the way for agents…
…The problem we faced at the time we released Mythos Preview was that no existing public exploit benchmarks were difficult enough to capture Mythos Preview’s capabilities in our initial testing. Over…
…In response to the model, one PRC cybersecurity analyst wrote that China is “still sharpening our swords while the other side has suddenly mounted a fully automatic Gatling gun.” Frontier AI capabilities…
…Doing so reveals patterns that challenge traditional assumptions about cybersecurity—for example, the level of risk a threat actor poses can be assessed via metrics like technical sophistication or breadth of techniques…
…In addition to human-initiated stops, agent -initiated stops are also an important form of oversight in deployed systems. On the most complex tasks, Claude Code stops to ask for clarification more…
To show you the most relevant results, we’ve omitted some entries very similar to those already shown. Repeat the search with the omitted results included.