A year ago, in How AI Is Rewriting the Rules of Cybersecurity, we looked at how advances in AI could bring in a new generation of multimodal cyberattacks, increasingly run by autonomous agents. That future is arriving fast.
Check Point Research found that in 2025 the average organization faced 1,968 attacks a week, up 18% from 2024. Fully agent-orchestrated attacks aren't the norm yet, but AI agents play a growing role in finding vulnerabilities.
CrowdStrike's 2026 Threat Hunting Report found that detection leads triggered by AI agents are growing 2.5 times faster than those triggered by humans.
The trend is clear, and it raises an awkward question for security teams: how do you test your defenses when your attacker is increasingly an AI agent?
French startup Fleuret AI is trying to answer it. Founded in January 2026 by childhood friends Yanis Grigy (CEO) and Augustin Ponsin (CTO), it launched its pentesting product in June. Despite its young age, the company just raised €4 million from RAISE Ventures, Auriga Cyber Ventures, Wind, Better Angle, Bpifrance and a group of cybersecurity-focused business angels.
