Anthropic's help page lists 16 Claude models back to Opus 4.5 as carrying its text watermark

Anthropic says it is adding the watermark to models released before 2 August 2026, which the EU law treats under a transition period. Neither of its pages gives a 3 November date.

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

Check our sources · 5 facts from 2 sources

Anthropic's Claude Help Center lists 16 models, among them Opus 4.5, Sonnet 4.5 and Haiku 4.5, as having first-party text watermarks, and says marking will apply to supported models wherever Claude is offered, worldwide.

The help article says Anthropic is adding watermarks to outputs from models released before 2 August 2026 and that existing models are in progress.

Anthropic's post of 14 August, updated on 1 September, says it is implementing watermarking to comply with the EU AI Act, and that the EU law includes a transition period for Anthropic models launched before 2 August 2026.

The post describes the watermark as a way of determining the likelihood that Claude was involved in writing a text, and says a detection API is in private preview for eligible organisations such as regulators, researchers and media groups.

Neither page mentions 3 November, so the date reported elsewhere for older models is unconfirmed here.

Your reaction

We count reactions per story and day, never who reacted. The counts help us choose what goes in the monthly issue. If you are signed in, your own page shows yours too.

Check our sources

Every sentence above is checked against these 2 sources.

1 How Claude marks AI-generated contentAnthropic (Claude Help Center) · 3 facts

Open the source archived copy

The Claude Help Center article How Claude marks AI-generated content lists as having first-party text watermarks: Fable 5.1, Fable 5, Mythos 5.1, Mythos 5, Opus 5.5, Opus 5, Opus 4.8, Opus 4.7, Opus 4.6, Opus 4.5, Sonnet 5.5, Sonnet 5, Sonnet 4.6, Sonnet 4.5, Haiku 5.5 and Haiku 4.5, which is 16 models.

Claude Opus 4.5
cite
The article says marking will apply to output from supported models wherever Claude is offered, worldwide.

wherever Claude is offered, worldwide
cite
The article says Anthropic is adding watermarks to outputs from models released before August 2, 2026, and that existing models are in progress; it does not mention 3 November.

Anthropic is adding watermarks to outputs from models released before August 2, 2026.
cite

2 How Claude's text watermarking worksAnthropic · 14 Aug 2026 · 2 facts

Open the source archived copy

Anthropic's post of 14 August 2026, updated 1 September 2026, says Anthropic is implementing watermarking to comply with the EU AI Act and that the EU law includes a transition period for Anthropic models launched before August 2, 2026; it does not mention 3 November.

The EU law includes a transition period for Anthropic models launched before August 2, 2026
cite
The post says watermarking is a way of determining the likelihood that Claude was involved in writing the text, and that a detection API is in private preview for eligible organisations such as regulators, researchers and media groups.

We are releasing a detection API in private preview.
cite

Topics

The morning email

On the mornings we publish: the three top stories and up to four short ones. Free.

We email you a link to confirm. An issue may include one sponsor, always labelled Sponsored · Advertisement. Our emails count opens and clicks, not who made them. Unsubscribe in one click. What we keep