A newly funded autonomous offensive security startup is raising the stakes for cyber defense in education and research environments. The company A—backed with $37 million including Lightspeed Venture Partners—deploys an AI-powered approach that continuously breaks into client systems to discover real attack paths and help remediate before human attackers strike. The reporting describes the company as emerging from stealth after Anthropic’s Mythos model showcased autonomous vulnerability discovery in earlier tests. In A’s early proof-of-concept, it allegedly surfaced 1.2 million exposed sensitive records—including Social Security numbers—left undetected for years by existing tools. Investors and security leaders are using the development to argue that the time gap between discovery and compromise is shrinking from attackers’ perspective. The funding round and product maturity highlight how fast AI can compress the offense-defense timeline. For higher education, where identity systems, research data, and student information are prime targets, the emergence increases pressure to update incident readiness, vulnerability management, and monitoring workflows.