Anthropic, Microsoft AI systems top cyber benchmarks (opens in new tab)

AI systems built for cybersecurity tasks posted several benchmark and product announcements this week. Anthropic’s Claude Mythos was described as the first model to clear all cyberattack simulations from Britain’s AI safety agency, while Microsoft disclosed MDASH, which GeekWire said topped Mythos on a cybersecurity benchmark. OpenAI unveiled Daybreak, reports described Mistral’s effort to serve banks, and other outlets linked the new tools to cybersecurity labor and bank remediation.Highligh...