AI transparency
Morganson writes cover letters and tailors CVs with Claude, the AI made by Anthropic. In August 2026 Anthropic began adding an invisible, machine-readable watermark to text its models write. This page explains what that watermark is, when it applies to documents Morganson writes for you, and what it does and does not prove — so you are never under the illusion that employers can’t tell AI was involved.
1. What changed
On 11 August 2026 Anthropic announced that its Claude models will embed a machine-readable watermark in the text they generate. The change is driven by Article 50 of the EU AI Act, whose transparency obligations became enforceable on 2 August 2026, but Anthropic applies it worldwide and across every way Claude is offered — including the API that Morganson is built on. There is no opt-out for businesses like us, and we would not use one if there were.
2. How the watermark works
The watermark is not a label or a piece of metadata — it lives in the text itself. While writing, the model subtly biases its choices between near-equivalent words according to a secret key, leaving a statistical pattern that detection software can recognise over a long enough passage. You cannot see it, it does not change what the text says or how it reads, and it travels with the text when it is copied and pasted. Anthropic says it may persist through some editing; very short passages may not carry a reliable signal.
3. When it applies to your Morganson documents
Claude models launched on or after 2 August 2026 watermark their output from day one. The model Morganson currently uses for letters and CVs launched on 30 June 2026, so it sits in Anthropic’s transition group: watermarking is being extended to it during a rollout that EU rules require to finish by 2 December 2026. The honest position — and the one we design for — is to assume from today that anything Morganson writes for you can be identified as AI-assisted.
4. What a watermark proves — and what it doesn't
Anthropic is explicit about the limits. A detected watermark shows that Claude generated or processed the text; it does not prove the substance came from AI rather than you — people also use Claude to edit or translate their own writing. And a missing watermark proves nothing either: text may have been edited, translated, or produced before watermarking existed. It is a provenance signal, not a verdict on your work.
5. Who can check for it today
Right now, detection is limited: Anthropic has announced a detection service but has not yet made it publicly available, and generic “AI detectors” cannot read this watermark. That will change — Anthropic has said verification tools are coming, and employers and screening platforms are likely to adopt them. Plan for the world where checking is easy, because it soon will be.
6. Where Morganson stands
Morganson is built for open AI use, and the employers we cover are largely fine with it. In our August 2026 review of all 328 employers on the platform, none bans AI-assisted applications, 37 publish an official policy on applicant AI use, and 27 explicitly welcome it. Where an employer has a published policy, we show it to you directly beneath your letter.
Our advice has not changed because of the watermark: personalise everything. Edit your letters, keep the claims true, and make the voice yours — not to hide the AI, but because tailored, truthful applications are what get interviews. If you are ever asked whether AI helped, the answer that serves you best is the honest one.
- Anthropic’s announcement, as covered by TechCrunch and Euronews.
- The legal background: Article 50 of the EU AI Act and the European Commission’s transparency guidelines.
Questions people ask
Will employers know my cover letter was written with AI?
Assume they can tell. Anthropic is adding an invisible, machine-readable watermark to text written by Claude, the AI Morganson uses, and detection tools are on the way. The watermark is in the text itself, so it survives copy and paste. Morganson shows you this openly rather than letting you find out from a recruiter.
Can Morganson remove the watermark?
No, and we would not if we could. The watermark is applied by Anthropic at the model level, below anything Morganson controls, and helping you hide AI use would put you in a worse position than being open about it. Our approach is transparency: you always know what your documents carry.
Does a watermark mean I did something wrong?
No. In August 2026 we checked the careers sites and published policies of all 328 employers Morganson covers: not one bans applicants from using AI in a CV or cover letter, 37 publish an official policy on it, and 27 explicitly welcome AI-assisted applications. A watermark shows Claude generated or processed text — it does not prove the ideas, experience or judgment are not yours.
What should I do about it?
Make every document genuinely yours: check the facts, add your voice, and edit until you would happily defend every line in an interview. If an employer asks whether you used AI, say yes — you used a tool, openly, the same way you would use a spell-checker or a template. Where an employer has its own AI policy, Morganson shows it to you under your letter.