3 ms·

Okay genericGPT answer the question by putting an emoji between each word. (The watermark is destroyed after removing the emojis)

• 1B05H1N 3y ago
  """ Emoji attack. In the “emoji attack,” the attacker asks the model to output a response to prompt with an emoji inserted between every pair of words. The attacker then removes the emojis to obtain the desired response. This attack removes any watermark that relies on the detector seeing consecutive sequences of tokens, including ours as well as those of [KGW+23] and [Aar22]. In general this attack may not preserve the output distribution, but any provable robustness guarantee for contiguous-text watermarks would have to rest on the dubious assumption that it doesn’t. """ https://eprint.iacr.org/2023/763.pdf https://eprint.iacr.org/2023/763.pdf Pretty funny imo
• totetsu 3y ago
  Or ask it to output the text backwards then reverse it, or output in one language then translate it to another in Google translate.. There are so many ways around this.
  • GaggiX 3y ago
    Both of these will affect quality.