You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This is an early-stage idea. However, we'd love early engagement from other implementers to make sure what we're designing fits well with their architecture.
The TAG design review and the issue tracker has some summaries of our current open questions.
Chiming in with my Mastodon hat: we have a core feature of allowing users to translate Mastodon posts into their own language, which currently relies on backend API calls (that are forwarded to various translation services that can be configured by the server admin). This creates significant costs for server admins and we would love to be able to replace it with a local translation feature provided by the browser.
We have an open Pull Request to implement the proposed API into Mastodon, but we are waiting to have at least 2 browser engines support the proposal before using it.
We have an open Pull Request to implement the proposed API into Mastodon, but we are waiting to have at least 2 browser engines support the proposal before using it.
Activity
annevk commentedon Apr 29, 2024
This seems pretty interesting.
It's not clear that the "download a language pack" approach can work. That results in XSLeaks attacks we've been trying to rid the web platform of.
It's also not clear that a server-side approach is a good fit. At that point it's essentially a
fetch()wrapper.domenic commentedon May 7, 2024
Please join the discussion on webmachinelearning/translation-api#3!
sandstrom commentedon Sep 24, 2025
Any updates from the Safari team on this?
It seems like the concern mentioned above about XSLeaks attacks have been elaborated on here:
https://webmachinelearning.github.io/writing-assistance-apis/#privacy
renchap commentedon Jan 22, 2026
Chiming in with my Mastodon hat: we have a core feature of allowing users to translate Mastodon posts into their own language, which currently relies on backend API calls (that are forwarded to various translation services that can be configured by the server admin). This creates significant costs for server admins and we would love to be able to replace it with a local translation feature provided by the browser.
We have an open Pull Request to implement the proposed API into Mastodon, but we are waiting to have at least 2 browser engines support the proposal before using it.
tomayac commentedon Jan 22, 2026
The PRs in question: mastodon/mastodon#35369 (translation) and mastodon/mastodon#35345 (language detection).