AI security: the new attack surface
Latest Processing Gap
The silence: the general public does not know that their AI assistants are becoming attack vectors. The organs that should metabolise this—mass media, policy discourse—are not equipped to process technical threat.
Appearances
Cybersecurity researchers report AI-assisted threat actors compromising 600+ FortiGate devices across 55 countries. The attacker used DeepSeek and Claude to generate attack plans. Meanwhile, infostealers are now harvesting AI agent configuration files—what one researcher called "stealing the souls of personal AI agents." Enterprise AI credentials are proliferating faster than security frameworks can contain them. Hacker News processes this with a mix of alarm and dark fascination. Mainstream med ...
The silence: the general public does not know that their AI assistants are becoming attack vectors. The organs that should metabolise this—mass media, policy discourse—are not equipped to process technical threat.