If anyone had heard in the 2010s the prediction that YouTube VP, Amjad Hanif, is quoted as saying in a Bloomberg article about AI dubbing, skepticism would have likely reigned on social media: in five years, “every video uploaded to YouTube could be dubbed automatically into every spoken language.” 

In 2025, this is more plausible, so no great skepticism, no millions of exploding-head emojis. Well, perhaps not every spoken language. But at the dizzying rate language AI is advancing (just look at what HeyGen’s Avatar or ElevenLabs’ Actor Mode can do), AI dubbing is set to improve and dominate over human dubbing a lot sooner than 2030.

Harder to predict are ethical and legal considerations for the public consumption of synthetic dubbing (and voice cloning). Yet, the technology, whether straight from YouTube or via a third-party AI dubbing provider, can be deeply transformative for organizations and individual creators. It is no longer just about reaching new global audiences, it is about improving access to content while guaranteeing accessibility and increasing viewer engagement.

We asked readers if they think that YouTube will manage to AI dub “all videos into every spoken language within 5 years,” and over half (53.2%) said “definitely not.” A little over a quarter (27.4%) of respondents think it is possible, and the rest (19.4%) are certain it will happen.

AI’s Legal Quickening

AI translation is pervasive, and although less openly acknowledged than other uses of AI, in regulated content, it is no longer operating behind the scenes. That much was evidenced by research conducted for Slator’s Legal Services and Language AI Report. 

While “legal translation” covers a wide variety of subfields, AI’s increasing speed and fluency have made its use seem like less of a liability. But as one professional legal translator commented for the report, “when you’re dealing with more technical legal documents, you need to be more careful with AI output. […] Junior lawyers use these tools indiscriminately.” 

It stands to reason, then, that legal AI translation should still be customarily edited by humans. In practical terms, however, law offices and in-house counsels might find that unedited AI translations suffice (in their view).

When asked if they had ever signed something they knew was AI-translated, over half of readers (51.1%) said they had not. A bit over a quarter of respondents (25.5%) would not know, while almost one in five (19.1%) would never sign [if they knew], and the rest (4.3%) have knowingly signed an AI-translated document.

AI Besties

Building new capabilities in-house, such as AI translation, could come at a high cost for traditional language services providers (LSPs). Beyond the design and proof of concept phases of a new offering lies product development, and that requires people and capital.

A way to find both, and have a chance at staying competitive, is through collaboration with a partner. And a partnership can take many different forms. Consider, for example, Phrase’s Q2 2025 partnerships, one with Acclaro and another with Clearly Local for co-funded roles. Or KUDO’s way, which welcomed Simultaneous On Air and Argo Translation as partners and resellers.

For some companies, whether they engage the partner for a fee or permanently connect to the partner venture’s destiny by way of equity or through split revenues on co-served accounts, it makes sense to bet on market relevance and competitiveness with a little help from a “friend.”

We asked readers if AI has made partnerships with friendly competitors more important, and the response chosen by almost half (48.3%) was “definitely yes.” Close to a third (27.6%) of readers are not sure about it, while one in five thinks it is probably so. For the rest (3.4%), the answer is no.

New Options for iOS Default Translation App

Apple began allowing translation apps other than its own Translate to be the default option for iPhones since it rolled out iOS 18.2 in December 2024. And going by the steps described on Apple’s developer site, it appears that integrating system-level translation capabilities into an existing app is a matter of just a few hours of work.

Among other basic requirements, to function as a default translation app, the application needs genuine translation functionality, meaning through a local model or a cloud API, along with a specific entitlement declaring its eligibility to be a default option.

Reverso Context and DeepL are two of the translation options already available to tick as default on the iPhone, with more likely to follow soon. We wanted to know if readers with iPhones had already selected a default translation app, which the vast majority (78.8%) had not done. About one in seven (15.2%) respondents have done so, and for the rest (6.0%), their preferred app was not listed as an option at the time of voting.