French Startup Raises €4 Million to Let AI Agents Hunt for Cybersecurity Weaknesses

Date: October 5, 2026

A French cybersecurity startup is using artificial intelligence to automate one of the most traditional jobs in cybersecurity: breaking into computer systems before real attackers can.

Paris-based Fleuret AI has raised €4 million in pre-seed funding to develop AI agents that continuously perform penetration testing. The round was led by RAISE Ventures, with participation from Auriga Cyber Ventures, Wind Capital, Better Angle and cybersecurity investors.

The company believes traditional penetration testing is becoming too slow for the modern software environment.

Companies change their systems constantly.

Security testing, meanwhile, is often performed only periodically.

Fleuret wants to make that process continuous.

Meet the AI Pentesters

Fleuret has developed two AI agents called Emile and Champollion.

The agents are designed to map a company’s environment, explore applications and APIs, identify vulnerabilities and attempt to exploit them.

Importantly, the system is designed to provide evidence that a vulnerability can actually be exploited.

That is different from simply identifying a suspicious piece of code.

A security team needs to know whether a weakness can realistically be turned into an attack.

Fleuret’s agents attempt to demonstrate that.

From Annual Audit to Continuous Testing

Traditional penetration testing often works like an annual health check.

A security company examines a system, produces a report and identifies vulnerabilities.

The problem is that software does not stop changing after the report is delivered.

A company might deploy a new application next week.

It could modify an API the following month.

A new vulnerability could then appear.

Fleuret argues that AI agents can continuously monitor the environment and launch new tests as the attack surface changes.

That could change the economics of cybersecurity.

Instead of paying for a large manual test once or twice a year, companies could potentially use automated AI testing continuously.

Europe Has a Particular Reason to Care

The company is explicitly positioning itself as a European cybersecurity business.

Fleuret says it wants to make offensive security accessible to companies that cannot afford frequent traditional penetration tests.

The regulatory environment also creates additional pressure.

European organizations are facing increasingly demanding cybersecurity requirements, including obligations associated with NIS2.

That means businesses need to demonstrate that their systems are adequately protected.

Automated testing could help organizations keep security assessments closer to the pace at which their infrastructure changes.

AI Against AI

There is also a bigger trend behind Fleuret.

Cybersecurity is becoming an AI-versus-AI competition.

Attackers can use AI to search for vulnerabilities faster.

Defenders can use AI to find those vulnerabilities first.

European defense technology company Thales has similarly argued that organizations need to use AI defensively as attackers increasingly use AI to accelerate cyber operations.

This could lead to a new generation of cybersecurity systems that continuously test, monitor and repair digital infrastructure.

Human security specialists would still be essential.

However, AI agents could increasingly perform the repetitive investigative work.

The Startup Already Has Customers

Fleuret says it has approximately ten employees and customers including Brevo, Stoïk and Yogosha.

The new funding will be used to hire additional people in AI, software engineering and offensive security.

The startup’s ambition is explicitly European.

That is notable because cybersecurity is becoming one of the strongest areas where European technology companies can compete globally.

The continent has large financial institutions, industrial companies and governments with strict security requirements.

Those customers need technology that can protect increasingly complex infrastructure.

What Happens When AI Finds a Vulnerability?

The most interesting part of Fleuret’s approach may ultimately be what happens after the AI discovers a weakness.

Finding a vulnerability is only the beginning.

Fleuret is building systems for prioritizing vulnerabilities according to exploitability, integrating findings with engineering tools and automatically testing whether a fix actually worked.

That creates a continuous feedback loop.

Find the problem. Prove it. Fix it. Test the fix. Repeat.

For modern software environments, that could be considerably more useful than a security report that becomes outdated weeks after it is delivered.