Skip to content

Ask HN: What if we provided support for AI guidelines at the kernel level?

1 pointPJHkorea1 comment
On HN

If the existing AI guideline approach is akin to giving a criminal (the AI) moral education (training) to encourage good behavior, how about creating a kernel-level switch that forcibly cuts off the electrical signals to its muscles the moment it grabs a weapon—that is, the moment it attempts a "jailbreak" via an adversarial method?

Comments

For instance, how about defining a set of values for the AI at the kernel level and matching vector addresses based on those values?

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.