Adversarial Attacks on LLMs

Lilian Weng · 25 Oct 2023

Read the piece on lilianweng.github.io ↗

Lilian Weng did not write this page.

Every claim below was made in this piece, quoted word for word and numbered in the order the piece makes them, so you can read it there rather than take our word for it. The sentence above each quote is our reading of the claim, not their wording. Each quote was checked against a stored copy of the page at build time; where the two differ, the quote is the fact.

Our reading
Universal adversarial trigger attacks are easy to detect because the learned trigger tokens tend to be nonsensical.

Their wordsOne drawback with UAT (Universal Adversarial Trigger) attacks is that it is easy to detect them because the learned triggers are often nonsensical.
Our reading
Higher-perplexity adversarial attack text is easier to detect and mitigate.

Their wordsHigh perplexity makes an attack more vulnerable to be detected and mitigated.

Keep reading

The other pieces are the ones whose claims are worded most like this one's. That measures language, not agreement: they may argue the opposite.