Innovation & Tech

AI translation enters live calls as Google expands to 63 languages

Google Meet is expanding live translation to 63 languages as Shopee video orders rose over 45%, while Chamet and other platforms embed AI translation inside calls and chats.

By Grace Kim

4 min read

Beyond language barriers: How AI is transforming real-time communication across platforms
Beyond language barriers: How AI is transforming real-time communication across platformsPexels / pixabay

What's News

  • Google Meet live speech translation is expanding to 63 languages through Gemini 3.5 Live Translate.
  • Southeast Asia’s digital economy is projected to reach US$368 billion GMV in 2026, up 17% year on year.
  • Shopee livestreaming and short-form video orders grew more than 45% year on year in Q3 2026.
  • Chamet’s real-time AI-translated subtitles support more than 50 languages.
  • Meta introduced Instagram Reels translation, dubbing and lip-syncing in October 2025; WhatsApp added on-device chat translation in September 2025.

Google Meet is expanding live speech translation to 63 languages through Gemini 3.5 Live Translate, Google said in an October 2026 product update. The feature detects the language being spoken, translates it into each participant’s preferred language in near real time and adds synchronised captions.

TNGlobal’s analysis frames that launch as part of a broader shift: AI translation is moving out of standalone apps and into chat threads, short videos, livestreams and video calls. Users can increasingly read, hear or watch content in another language without switching platforms.

What makes translation a business issue in Southeast Asia?

Google, Temasek and Bain & Company project Southeast Asia’s digital economy to reach US$368 billion in gross merchandise value in 2026, up 17% year on year, according to the e-Conomy SEA 2026 report.

Sea Limited reported that livestreaming and short-form video orders on Shopee grew by more than 45% year on year in the third quarter of 2026. TNGlobal said language can become a barrier as consumers discover products and interact with sellers through video.

TNGlobal links the wider shift to cross-border connection, commerce and collaboration. Its practical conclusion is simple: translation now sits where buying and conversations already happen, not in a separate utility.

Why does trust become harder when translation is embedded?

The UN Office on Drugs and Crime has warned that AI helps criminal groups scale multilingual social-engineering and phishing operations. Generative AI can produce persuasive messages in different languages and support simultaneous conversations with victims, according to UNODC.

TNGlobal says that combination weakens a familiar warning sign: awkward phrasing and poor translation are less reliable indicators of deception. Language alone may no longer tell users whether a message or sender is trustworthy.

Accuracy, privacy and user control remain central, TNGlobal says. Users need clarity on when AI produced or translated content and how platforms use their data, while translation tools must preserve meaning, tone and context.

TNGlobal does not argue that AI translation is inherently risky. It says platforms should assess the wider context of translated conversations and give users safety tools when something feels wrong. Platforms must pair convenience with context-aware protections.

Which platforms are putting translation inside conversations?

The product releases cited by TNGlobal place translation in three layers of digital activity: meetings, social video and messaging.

  • Google: In its October 2026 update, Google said Meet was expanding live speech translation to 63 languages through Gemini 3.5 Live Translate. Automatic language detection selects each participant’s preferred language, with near-real-time translation and synchronised captions.
  • Meta: In October 2025, Meta said it was using AI to translate, dub and lip-sync Instagram Reels, helping creators reach audiences who speak different languages.
  • WhatsApp: In September 2025, WhatsApp said it had introduced message translation within individual and group chats, with translations processed on users’ devices.

TNGlobal says the change is particularly consequential on video-first social platforms. In a live exchange, users do more than read translated text; they interact with another person, where context and trust matter.

TNGlobal’s conclusion is straightforward: translation now runs inside the chat, reel or call rather than in a separate application.

How is Chamet pairing translation with safeguards?

Chamet, a video-first social platform, says its real-time AI-translated subtitles support more than 50 languages. Users can follow conversations without opening a separate translation tool.

David Qi, Chief Marketing Officer of Chamet, said: “We see that translation changes what people expect from a conversation with someone on the other side of the world. People expect to be understood, and they expect the platform to take their safety seriously,” and added: “We see translation and safety as two parts of the same job.”

Chamet combines multimodal AI moderation and suspected-minor detection with human review, TNGlobal said. The platform also offers real-person face verification, in-app reporting and blocking tools, and an appeals process.

Chamet sets four operating principles for platforms that place translation inside a conversation:

  • Know who is on the call. Verify people, not just accounts.
  • Keep humans in the loop. AI flags. People decide.
  • Say what is machine-made. Label translated content.
  • Give users an exit. Make report, block and appeal easy to find.

TNGlobal says these safeguards do not remove every risk. They help keep easier communication from coming at the expense of user confidence.

What will decide the next phase?

TNGlobal expects AI translation to keep evolving from a standalone utility into an integrated layer across real-time communication. The report says the larger contest will be user confidence: platforms must improve understanding while maintaining trust and safety.

The practical test for platforms embedding translation in live calls, chats and commerce is straightforward: language coverage matters, but trust and safety controls will decide whether users stay.

Source: technode.global

Share this article:

More from Grace Kim

Grace Kim

Show full bio

Market editor covering people, launches and day-to-day practice in entrepreneurship, startups and venture funding at Business Bearings.

2,487 articles

Related articles

« Previous articleNext article »