Anthropic understandably being vague about the methods, but watermarking LLM text can be done at a syntactical level - eg it survives rewriting because it’s a pattern in the word choice. More details on methods here (it’s really interesting) www.seangoedecke.com/text-ai-wate...
Who's Who of AI
James Vincent
Why they matter
Directory member with public evidence across Evaluation & benchmarks.
- AI signals
- 1 past 30d
- Sources
- 1 distinct domains
- Discussions
- 0 past 30d
- Latest signal
- 7d ago
journalist and writer. author of BEYOND MEASURE, a history of measurement; a New Yorker, Economist, Times book of the year. former senior editor at The Verge. buy my book here: https://linktr.ee/BeyondMeasureBook. website: jamesvincent.info
What they're sharing
Text AI watermarks will always be trivial to remove seangoedecke.com
Articles & links
Their own posts
Recent commentary
had a conversation with a writer friend and we concluded that the worst thing that could happen as a result of AI text generation - the very worst - is if people lose the will to write; if we begin to believe that there is simply no need, no reason to try, and elect to stay mute
how miserable will life be when instead of showing you idealized versions of other people, adverts use AI to show you idealized versions of yourself — scrolling past your face, but more beautiful; your head, on a more attractive body... hopefully the suicides will force standards
Their network
In James Vincent's orbit
Center = James Vincent. Left = members they follow (green edges). Right = members who follow them (blue edges). Top = mutual follows (orange edges, slightly larger). Drag any node to reposition; click to open that profile.