Hackers Are Using AI to Write Exploits and Automate Attacks at Scale

The same AI tools marketed to security teams for vulnerability research are being used by attackers to write exploits and automate attack delivery at scale. This was not a surprise to anyone with access to the tools. The tools were marketed as dual-use from the beginning. The restriction was assumed to be social, not technical.

Cybersecurity departments briefed on this in advance. Adequate attended none of these briefings. The decision to distribute unrestricted access to foundation models was made with full knowledge that attackers would acquire them through standard channels. The timeline between 'here are tools for defense' and 'attackers are using these tools for offense' was measured in weeks.

Scale is now the constraint that has been solved. Individual attacks required individual effort. Now attackers automate against thousands of targets simultaneously using the same models that find the vulnerabilities. The briefing recommended certain mitigations. The briefing was not attended. The slide deck still exists.