ZSZainSpireLibrary
Career guide · updated 13 Nov 2026

Claude's watermark and your resume: the facts, and the honest way

Since August 2026, text from newer Claude models carries an invisible watermark because of the EU AI Act. Reels are scaring people into "removing" it. Here's what's actually true, and how to use Claude so your resume is genuinely yours.

The 3 facts 1. It isn't hidden characters. It's a statistical pattern in Claude's word choices, so nothing extra is "inside" your text.
2. HR can't just check it. Detection is in private preview for approved organisations (regulators, researchers, media, fact-checkers and similar), not the public or employers in general.
3. A detected mark isn't proof. Anthropic says it only means Claude may have processed the text, for example proofreading or translating. Heavy editing, short text and older models can also mean no mark.

01Don't install random "watermark remover" skills

A skill is a set of instructions, and sometimes scripts, that you hand to Claude. Uploading one from an unknown GitHub account means trusting a stranger with what Claude runs on your files. Most claim to strip "hidden characters", which isn't how the watermark works anyway.

02Use Claude as a coach, not a ghostwriter

Copy this prompt, fill the brackets, and answer Claude's questions with real facts:

You are a senior recruiter for [target role] roles in [country/industry].
Here is my current resume and the job description.

RESUME:
[paste your resume]

JOB DESCRIPTION:
[paste the job ad]

1. Tell me what's missing for this job, in a short list.
2. Suggest a better structure (sections + order).
3. For each bullet, ask me ONE question that would help me add a real number, tool or result.
Don't write new bullets for me and don't invent anything. I'll answer, then you help me tighten MY wording.

03Make every line yours

04Final checklist

Source Claude Help Center: How Claude marks AI-generated content