الوئام
الوئام
Ready to play
Ready to play
Anthropik has announced the release of a new technique to add an invisible statistical fingerprint to AI model outputs, such as Claude, in order to comply with European transparency laws. This technique relies on selecting words in a way that repeats across texts, creating a pattern that can be detected invisibly without affecting the quality of the writing or the system's speed. The fingerprint's effectiveness is limited to longer texts or those with multiple possibilities, and it diminishes with shorter content or responses requiring specific answers. It is not significantly affected by human corrections or translations but can be removed through comprehensive rephrasing. The technique does not guarantee full ownership proof of the text or identify the author. It is currently undergoing testing and development to provide detection tools based on private keys, as part of the company's commitment to European transparency regulations.
Notice: This Is an AI-Generated Summary
Comments (0)