at://
/
app.bsky.feed.post
/
3m64dghvs722r
sign in
All
4
Record
2
Post
1
PostEmbed
1
Post
by @danabra.mov
PostEmbed
by @danabra.mov
Record
by @jimpick.com
Record
by @atsui.org
+ new component
Post
You can sabotage AI with poetry, pass it on
6mo
Nathan K. Hensley
Looks like LLMs are *very* vulnerable to attack via poetic allusion: "curated poetic prompts yielded high attack-success rates (ASR), with some providers exceeding 90% ..." https://arxiv.org/html/2511.15304v1
