Hi folks!
We had a great first-ever translation AI community event at Tech Week in San Francisco on Thursday evening, in person and off the record, with panelists and moderators from Adobe, Google, ModelFront, DeepL, Microsoft, Meta, Twitch, Disney, Krisp and LocWorld.
Translation features keep popping up in the tools people already use. DeepL now works inside Copilot, ChatGPT and Claude through MCP, and Vodafone and Ericsson are testing translation built into the mobile network itself, with no app needed.
For enterprise localization teams, there’s now an open TMS built for AI. Until now, MT and AI in general were an afterthought in translation management systems. Now ModelFront has announced the launch of OpenLoc. The code will be released on GitHub on Monday under BSL 1.1.
Enjoy — and share!
🗞️ News
The first-ever translation AI event at Tech Week.
Vodafone and Ericsson trial network-level translation of regular voice calls between English, Spanish, Italian, French and German, on any phone and with no app.
ModelFront announces OpenLoc, an open TMS built for AI that works with any LLM and can be managed through MCP.
The European Commission’s eTranslation passes 1 billion requests in 2026, more than in any previous full year.
Audible quadrupled English translations of European audiobooks this year, mainly from Germany, France, Spain and Italy, all done by human translators.
🚀 Launches
DeepL launches an MCP server that lets Microsoft Copilot, ChatGPT and Claude translate text, documents and images using your organization’s glossaries and style rules.
Google launches offline mode for Live Translate on Pixel 9, 10 and 11, so conversations between English and 10 languages work without an internet connection.
Google Ads launches a beta AI campaign localization tool that translates English Search campaigns into 10 languages.
Zendesk launches real-time voice translation for contact centers in 13 languages, starting with closed early access.
Bilibili launches the open Index-Translate model family for 150 languages, with a free API and versions for long documents, dubbing and speech-to-speech translation.
UGREEN launches ClipBuds Pro2 earbuds, which translate in real time in 131 languages.
🛠️ Community launches
EdgeTranslator, an open-source desktop translator that runs Gemma 4 fully offline in 35 languages, with voice and image input.
noentenc, an open-source Python library for language detection and machine translation on CPU, without PyTorch.
DuskTranslate, a workspace for translating Japanese light novels chapter by chapter, with EPUB import and export and a glossary for each project.
aitranslator.cloud, a set of browser-based tools for technical manuals, with a TMX aligner and editor, and glossaries with forbidden terms.
NagaTranslate, a free translator for Naga languages from Northeast India, now with 4 more in beta: Tenyidie, Tangkhul, Rongmei and Phom.
📚 Research
Amazon opens a call for research proposals on multilingual and culturally contextual AI evaluation, including translation for low-resource languages (deadline: 4 November).
🎙️ Podcasts
SlatorPod on why ModelFront is launching an AI-native TMS, with CEO Adam Bittlingmayer.
❔ Questions
What is the difference between LLM and neural machine translation?
Has anyone tried JEV, Laya or other “System 1” models for quality estimation, as a faster alternative to LLM-as-a-judge?
Where can I find Arabic–English code-switched meeting audio with transcripts, for evaluating speech recognition?
Which Windows translator apps do people recommend as an alternative to RTranslator on Android?
📆 Events
WMT 2026 — 28 - 29 October — Budapest 🇭🇺
💻 Jobs
Founding Engineer at Awtomated — Berlin 🇩🇪
Localization Lead, Engineering and Automation at Apertera — Sofia 🇧🇬
Localization Engineer (with language skills) at McAfee — Cork 🇮🇪