Infohazard автор Nirro

Being halfway through an article only to find out it was not written by anyone is disheartening. Sometimes the signs are all over the place but we cannot simply remain constantly alert. This extension makes those signs (at least the ones that are easy to spot by software means) very obvious.

Telling LLM slop apart from genuine content is not a deterministic problem. It can never be done 100% reliably, so our approach is just to ensure you are not caught off guard when the signs start piling up. One em dash in an article might be a grammar nerd, but fifteen obvious telltales probably indicate that nobody made the effort to write what they are putting you through the effort of reading.

The visual clues are orange highlights on the suspect text and a small orange-black striped rectangle before the suspicious section starts. Hovering this rectangle, you can see what category it was flagged under inside a tooltip.

The categories are all derived from the "Signs of AI Writing" Wikipedia page.

Note that all processing happens inside your browser, we do not collect any data whatsoever, we do not send any data to third parties. This extension is Free software published under the GNU Affero General Public License v3.0 and it will always remain free to use and respectful of user rights.