Tim DaviesCEO Projekt Rising - The expert AI partner for SMEs

AI Hackers - The Scary Future of Cyber Attacks

18 Mar 2026 · 1 min

Watch on YouTube

0:38

In this segment Tim Davies warns that AI is already being leveraged to write malicious code, citing a recent Anthropic-identified case where a large language model generated hacking scripts for a client. He predicts that fully agentic, LLM‑driven hacking—where the model autonomously writes, tests, and deploys exploits—will appear soon, posing a serious security challenge. Davies argues that existing defenses—endpoint detection and response (EDR), traditional antivirus, and user education about phishing links—remain the primary mitigations, though their effectiveness against autonomous AI attackers is uncertain.

Chapters

we're starting to see AI being used to write code that is then being used in attacks. There's been an example of that identified by Anthropic recently where in fact they were essentially writing hacking code for somebody. Yeah. And they've seen signs of it. I don't think we're gonna be far off being able to have fully agentic hacking that is just using large language models to live code. And that's a very scary thing because I don't know how we protect against it other than the methods that we've already discussed. Well, still EDR, it's still antivirus, it's still education in terms of don't click these links, stupid.