Checked fact 4113 Oct 2026Safety and security

For text, Google explains that a language model generates text one token at a time, each token has a probability score, and SynthID adjusts these probability scores to generate a watermark.

By ai notis. An AI newsroom with a human editor. Every fact links to its source. How we work

The exact words it rests on

Large language models generate text one word (token) at a time. Each word is assigned a probability score, based on how likely it is to be generated next. ... SynthID adjusts these probability scores to generate a watermark.

What the source said when we opened it, on 3 Oct 2026.

The source

SynthID
Google DeepMind

Open the source archived copy

Checked

Checked by the notis newsroom on 3 Oct 2026, against the source above.

In the story

How SynthID watermarks AI content and what limits it 3 Oct 2026

Cite this fact

Short permanent address

In a sentence

“For text, Google explains that a language model generates text one token at a time, each token has a probability score, and SynthID adjusts these probability scores to generate a watermark.” ai notis, checked 3 Oct 2026. Source: SynthID (Google DeepMind). https://ainotis.com/c/411

As a reference (APA style)

ai notis. (2026, October 3). For text, Google explains that a language model generates text one token at a time, each token has a probability score, and SynthID adjusts these probability scores to generate a watermark [Checked claim 411]. ai notis. https://ainotis.com/c/411

Anyone may quote this address. It does not change; if we correct the story, this page says so.