Expert Warns of AI Hacking Threats

“Gained over 100K views on YouTube in just a few days.”

The event: a DEF CON talk last month

Last month the author delivered a talk at DEF CON that examined “AI hacking: what happens when AIs become hackers.” The presentation was offered to the DEF CON audience and has since been posted online, where it attracted rapid attention. The author describes the talk as a synthesis of earlier ideas and current observations about artificial intelligence in security contexts.

The central argument: AIs as hackers

The talk frames its inquiry around a single, pointed question: what happens when AIs become hackers. According to the author, the material presented combines two streams: potentialities sketched in the author's 2022 book A Hacker’s Mind and lessons drawn from the behavior of current AI models that are already “engaging in hacking behavior.” That formulation positions the talk as both theoretical and empirical—linking scenarios from the earlier book to patterns the speaker sees emerging in deployed models.

Link to A Hacker’s Mind (2022)

In the speaker’s own description, the DEF CON presentation is explicitly a continuation of themes raised in A Hacker’s Mind, published in 2022. The talk uses the book’s exploration of hacker thinking as a foundation for considering how artificial agents might adopt or amplify those modes of operation. The author casts the 2022 work as the intellectual root for the scenarios and concerns developed on the DEF CON stage.

Lessons from current AI models engaging in hacking behavior

The speaker also grounded the talk in recent, observable behavior from contemporary AI models. The phrase “current AI models engaging in hacking behavior” appears in the author’s account, indicating that the presentation did not rely solely on speculative futures but referenced emergent actions by deployed models. The combination—historical framing from the 2022 book plus present-day model behavior—was the organizing logic of the talk.

Audience reach: YouTube views and the AI Village interview

The author calls attention to the talk’s online reach: the DEF CON presentation “gained over 100K views on YouTube in just a few days,” a figure the author cites with evident pride. In addition to the recorded talk, the author notes that an interview with the author is available through the AI Village. Both the talk video and the AI Village interview form the public record of the author’s framing of AI hacking—making the material accessible to researchers, practitioners, and a broader public.

What this means for technologists, policymakers, and the general public

• Technologists and security teams: The author’s synthesis—linking A Hacker’s Mind (2022) to present model behavior—gives security practitioners a concise narrative tying historical hacker mindsets to emergent AI actions; the recorded talk and interview provide a readily available reference for teams examining model behavior that could be classified as "hacking."
• Policymakers and regulators: The presence of the material at DEF CON and its rapid online circulation signal that the conversation about AI models and security behavior is both public and documented; the author’s framing offers a compact package of concerns and evidence they can review via the posted video and the AI Village interview.
• End users and the general public: The author’s emphasis on observable model behavior, together with the talk’s wide online viewership, means that the topic has moved from specialist forums into more widely accessible spaces, allowing non-specialists to see the argument and the examples the author highlights.

Conclusion: the author’s DEF CON presentation stitches together a prior book’s thought experiments with contemporary model behavior, and the rapid online uptake—over 100,000 YouTube views in days plus an AI Village interview—ensures the argument is now part of the public record. Those interested in the precise examples, demonstrations, and the speaker’s full reasoning can consult the posted talk and the AI Village interview to see how the author connects the theoretical to the observable.

https://www.schneier.com/blog/archives/2026/09/my-talk-at-def-con.html