Doit-on se préparer à une avalanche de failles découvertes dans nos systèmes par les IA ?

Doit-on se préparer à une avalanche de failles découvertes dans nos systèmes par les IA ?

Should we prepare for an avalanche of vulnerabilities discovered in our systems by AIs?

by RISKINTEL MEDIA · View the channel on PSI

Applied Sciences & Engineering AI — News & Applications UYComputer scienceUYQArtificial intelligence (AI)
🎙 RISKINTEL MEDIA 👥 209K 📅 September 9, 2026 ⏱ 42 min 👁 5 📄 expert opinion 🧭 2026-09-09
Available in: English (current) Français

Keywords

AIvulnerabilitycybersecuritySOCpatch

Summary

In this interview, two cybersecurity experts, Laurent Bessé (Director of Cybersecurity at I Tracing) and Vincent Rois (Security Advisor at Splunk/Cisco), discuss the growing threat of AI-discovered vulnerabilities and the challenges of defending against them. They highlight that AI tools like Mythos can find vulnerabilities that humans might miss, and that the time to exploit vulnerabilities has dropped to under 7 days, sometimes under 24 hours. They argue that the traditional approach of prioritizing patches based on criticality is being upended because AI can chain multiple low-severity vulnerabilities into a critical exploit. They advocate for a ‘war of movement’ strategy, focusing on observability, data collection, and virtual patching. They also discuss the importance of integrating AI into secure development practices, as seen in Cisco’s Glasswing project, and the need for crisis management preparation. The conversation touches on the use of AI agents in security operations, emphasizing the need for hybrid approaches and careful oversight. The experts stress that while AI can accelerate both attack and defense, organizations must adapt their security paradigms to cope with the increasing volume and complexity of threats.

182 words

Critical Evaluation

Value of the Information & Strength of the Argument

The value of the information is high for practitioners, as it provides practical insights into how AI is changing vulnerability management and SOC operations. The experts argue convincingly that the traditional patch prioritization model is becoming obsolete, and they offer concrete alternatives such as virtual patching and enhanced observability. They also highlight the importance of integrating AI into secure development and the need for crisis management preparation. The argumentation is solid, grounded in real-world examples like the Glasswing project and Log4j, and they acknowledge the complexity and uncertainties involved.

Scientific Rigor, Source Quality, Title Accuracy

The discussion is rigorous in its technical depth, but the sources are not formally cited. The experts reference industry reports (M-Trends) and internal projects (Glasswing) without providing specific URLs or data. The title accurately reflects the content, which focuses on the potential avalanche of AI-discovered vulnerabilities and the need for proactive defense. The lack of formal citations and the low viewership limit the verifiability of the claims, but the expertise of the speakers lends credibility to the discussion.

182 words

Title / Content Match

The title accurately reflects the core topic: the potential surge in vulnerabilities discovered by AI and the implications for security teams.

Quality & Reliability

7/10

The discussion features two experienced cybersecurity professionals (director of cybersecurity at I Tracing, security advisor at Splunk/Cisco) providing informed opinions on the impact of AI on vulnerability discovery and exploitation. They reference concrete examples (Mythos, Glasswing project, Log4j) and industry reports (M-Trends). However, the conversation is largely anecdotal and lacks detailed citations or verifiable data, and the low view count suggests limited external validation.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

Concurring Sources

  • M-Trends 2026 — Referenced in the video as a report by Mandiant estimating time-to-exploit.

Contribution & Novelties

The interview provides a timely expert perspective on the impact of AI on vulnerability discovery and exploitation, emphasizing the shift from reactive patching to proactive defense strategies like virtual patching and enhanced observability. It also highlights the importance of integrating AI into secure development and the need for crisis management preparation.

Pour aller plus loin :

  • Mythos AI — Background on the AI tool mentioned.
  • Virtual patching — Explanation of the technique discussed.
  • Log4j vulnerability — Example of a widespread vulnerability mentioned.

82 words

Radar Profile

The radar profile shows high scores in information quantity and technical level, reflecting the depth of the discussion. The quality and reliability scores are moderate, indicating that while the content is expert-driven, it lacks formal citations and verifiable data.

Reliability 6/10