| 15:01 | <Jake Archibald> | Chrome folks: Are there regional variations in the model used by the Prompt API? Do users in the UK get the same model as users in China etc etc? |
| 16:37 | <smaug> | Btw, Thursday is May Eve. It may affect participation. |
| 17:29 | <jarhar> | I think Mike Wasserman works on this? |
| 20:45 | <Noam Rosenthal> | you mean apart from the requested language? I don't works on this but https://github.com/webmachinelearning/prompt-api/issues/175 comes to mind as this might be a fingerprinting thing. |
| 21:30 | <Jake Archibald> | I'm more wondering if, if a user is in China, they'd have a model that self-censors in ways models given to other users do not. Meaning, asking the Prompt API about Tiananmen Square would be a fingerprinting vector |