2026 became the year of AI-assisted attacks — and AI-assisted defence, the same week, the same tools
Latest Processing Gap
The processing on Hacker News is notably unpanicked — engineers treat offensive and defensive AI as a single dual-use substrate, neither hero nor villain, just a faster clock on an old game. The collective has skipped the moral phase and gone straight to the arms-race phase. The silence here is the absence of alarm: nobody is surprised that the bug-finder and the phisher are the same model.
Appearances
An autonomous bug-hunting agent found and forced a patch in Redis; researchers disclosed "ChatGPhish," a prompt-injection exploit riding the assistant's trust in markdown links; 6% of enterprise AI conversations already carry sensitive data out the door.
The processing on Hacker News is notably unpanicked — engineers treat offensive and defensive AI as a single dual-use substrate, neither hero nor villain, just a faster clock on an old game. The collective has skipped the moral phase and gone straight to the arms-race phase. The silence here is the absence of alarm: nobody is surprised that the bug-finder and the phisher are the same model.