Mock cyber attacks used to cost about $30,000, take two weeks, and tell companies what their defenses looked like six months ago. Then Anthropic’s Mythos showed up, exposing thousands of zero-day vulnerabilities.
A (yes, as in the letter) uses AI to continuously break into its own customers’ systems—finding real attack paths and fixing them—before actual hackers get the chance. The New York-based autonomous offensive security startup, emerged from stealth with $37 million in funding from Lightspeed Venture Partners, Cyberstarts, and angels including Wiz CEO Assaf Rapaport and Cyera CEO Yotam Segev, Fortune learned exclusively.
The timing is hard to miss. Anthropic’s Mythos model—unveiled April 8 in a restricted release—autonomously surfaced previously unknown security flaws across every major operating system and browser. This handed unsophisticated attackers a preview of what’s coming. “Today every teenager, every grandmother can take a laptop, access those frontier models, and do things that were once reserved for nation-state actors,” A CEO and cofounder Yossi Torati told me. “We will face a meltdown in the industry.”
Torati spent six years at Israeli incident response firm Sygnia helping organizations investigate and recover from major hacks and responding to breaches across roughly 40 countries. His cofounders, Omer Gull and Yuval Itzchakov, come from similar cyber backgrounds from Check Point, Hunters, and IDF Unit 8200.
