Penetration testing used to mean human experts, tools and long reports. A Security is pitching a different answer for a different attacker: autonomous AI agents that scan, chain vulnerabilities and look for a path inside.
The Israeli company emerged from stealth on June 8 with $37 million in funding and a blunt thesis. If attackers are using AI agents, defenders need AI agents that find and close paths first.
Two rounds, and a telling angel list
The funding came in two pieces: a $5 million seed round led by Cyberstarts, then a $32 million follow-on round led by Lightspeed and Cyberstarts five months later.
The angel list is the stronger signal. Wiz CEO Assaf Rappaport, Cyera CEO Yotam Segev and Dazz founder Merav Bahat invested personally. In cloud security, operators do not casually endorse another team with their own money.
Attack paths, not vulnerability lists
A Security calls the threat “weaponized AI”: autonomous agents that can industrialize hacking work. Its platform runs offensive and defensive agents together.
- The offensive side scans continuously and links separate weaknesses into a full attack path.
- The defensive side runs bounded tests, records the audit trail, verifies whether a finding is exploitable, and can adjust protections automatically.
The company is trying to move customers from a long list of possible vulnerabilities to a map of what can actually be used against them, plus remediation.
Why now
The timing is simple. Attackers can use AI to automate scanning, probing and chaining at a pace that manual defensive workflows cannot match. A Security is betting that automated attacks will force automated validation and repair.
The real test is whether its agents can survive production systems and outpace criminal tooling. For now, the funding shows that security investors expect the battlefield to shift from human-versus-human to agent-versus-agent.
Sources：SecurityWeek, GlobeNewswire and Calcalist / CTech reports on A Security；CocoLoop