Post #4167843
2026-07-28 15:55 UTC
@mttaggart@infosec.exchange That is a good checkpoint to have, but much as I love to have my "is this actually as big a deal as the hype suggest?" priors validated, I also want to remain epistemically humble. a plausible explanation for this number being so low is that right now defenders have the advantage simply because most attackers haven't integrated the open weights models into their toolchains yet; it's plausible to me that we are still in the "huh why is the water level so low" part of the tsunami
Replies (1)
-
@glyph@mastodon.social 2026-07-28 15:56
@mttaggart@infosec.exchange last year IIRC the state of the art is that the CCP would send an email to Anthropic asking for permission to do their world-spanning cyber operations and sometimes Anthropic would say "no" and close their account https://apnews.com/article/ai-cyber-china-hacking-artificial-intelligence-anthropic-4e7e5b1a7df946169c72c1df58f90295