Stable writing habits in model output can point to its source.
An LLM style fingerprint is like your dad in the family group chat. He may try new emojis, but he still ends every text with “OK.”
People use it to trace AI text and check who wrote it. It can be wrong. It can also raise privacy fights.
LLM
A style fingerprint comes from habits that repeat in LLM output.
AI detector
An AI detector may use these style clues to judge the source.
Content provenance
It can add source clues, but it is not as clear as a watermark.
Writing Style Cloning
Writing Style Cloning can copy the fingerprint and make the source harder to tell.