//
sign in
Post
by @danabra.mov
PostEmbed
by @danabra.mov
Record
by @jimpick.com
Record
by @atsui.org
+ new component
Post
Say “AI guardrails” again, I dare you, I double dare you!
3h
Alt: Say It Again Cereal - Pulp Fiction Scene where Samuel L Jackson threaten a dude with a gun
static.klipy.com
Say It Again Cereal - Pulp Fiction Scene
ddɐ˙ʎʞsʞɔɐlq˙uǝʌs@
"A new proof shows that a fixed set of guardrails placed on AI is not universally robust against adaptive adversarial prompts" www.nist.gov/news-events/...
6h
Adolfo Neto
« You can’t escape Gödel in math, and in AI, you […] can’t patch an [LLM] and then expect to be OK forever. You have to [constantly] search for weaknesses […] The goal is [to] make it financially prohibitive for attackers to […] break your AI system. [T]hat’s the cost of even partial security […] »
8h
The proof extends to AI the logic used by famed mathematician Kurt Gödel, whose incompleteness theorems have had a profound effect on math for nearly a century.
www.nist.gov
NIST Mathematical Proof Supports Transition to a Continuous-Monitor-and-Update Security Model for AI Systems
ddɐ˙ʎʞsʞɔɐlq˙uǝʌs@