Michaël Marcozzi

@marcozzi.net

🇪🇺🇧🇪 Permanent researcher in cybersecurity (fuzzing). Works at CEA List Institute from Université Paris-Saclay (France). http://www.marcozzi.net I make software safer by viciously torturing it to reveal its flaws. I can be kind too.

🎉 Our paper got accepted at #ASE2026! We protect open-source stacks from backdoor injections with fast, precise, fuzzing-based detection right in the CI & release pipelines. 🛡️ 👏 Congrats to my PhD student Dimitri Kokkonis, w/ Stefano Zacchiroli. 📄 Preprint and artefact coming soon!

Bild

Check out our ROSARUM benchmark, part of our #icse2025 paper on backdoors and fuzzing: ▶️ It is a fuzzing benchmark (can fuzzers trigger backdoors reliably and fast?) ▶️ It is a backdoor detection benchmark (can code analyses find backdoors reliably?) ⬇️ github.com/binsec/rosarum

GitHub - binsec/rosarum: A novel backdoor detection benchmark

A novel backdoor detection benchmark. Contribute to binsec/rosarum development by creating an account on GitHub.

github.com

Happy to read such enthusiastic reactions to our @icseconf.bsky.social paper! ☺️ "ROSA Sets a New Standard for Backdoor Detection." "If you work in cybersecurity, penetration testing, or software auditing, ROSA is a must-try in the fight against hidden threats." medium.com/@itsissachar...

ROSA: A Breakthrough in Backdoor Detection with Fuzzing

The security industry is constantly battling against hidden backdoors—maliciously planted vulnerabilities that grant attackers undocumented…

medium.com

🥳📰 Very happy and proud that our paper on finding backdoors with fuzzing was accepted at the main track of @icseconf.bsky.social! More details to follow soon 🙂 Congratulations and thank you to my students Dimitri Kokkonis and Emilien Decoux and co-supervisor Stefano Zacchiroli!

Bild