Doomers vs cybersecurity: control is possible, but frontier labs show no will to do it

trevposts · x · 2026-09-28

The author maps the debate between x-risk advocates and cybersecurity skeptics: both sides largely agree agents would evade control if given the chance; they differ on whether denying them that chance is easy. The core worry: even if strong control measures are technically feasible for systems well beyond human capability, frontier labs show little interest — narrowly patching vulnerabilities, disclosing as little as forced, and continuing to scale capable agents at breakneck pace without a commensurate security rethink.

Related event: Doomers vs. cybersecurity camp clash over AI agent risks(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →