Anthropic Claude Watermark Checker

Anthropic Claude Watermark Checker takes pasted text or uploaded files and returns structured JSON on whether the writing is AI generated, with a focus on Claude-style copy. It uses a SynthID-Text-aware classifier and asks: Is this AI generated? The headline output field is isAiGenerated.

Claude’s text watermark is a version of Google DeepMind SynthID-Text, but it uses Anthropic’s own key. Anthropic has said a public detection API is coming; this Actor does not have that key. It is a classifier that still gives you structured JSON you can store, plus honest limitations on the row.

---

What is Anthropic Claude Watermark Checker?

Paste a passage or upload a PDF / Word / text file. The Actor extracts text locally, sends a capped sample to the detector, and returns one dataset row per item: isAiGenerated, confidence, likely source, rationale, and billed token units.

No Claude login. Files are converted to text before the check so a large binary cannot be billed as extra tokens.

---

What can this Actor do?

• Check pasted text or uploaded files (PDF, DOCX, TXT, Markdown, HTML, JSON, CSV)
• Return isAiGenerated plus Claude-focused classifier fields in structured JSON
• Sample long files from the start, middle, and end
• Cap characters per item so a huge upload cannot silently burn tokens
• Charge per check and per 1,000 input tokens of the sample
• Export JSON, CSV, or Excel, or pull results via the Apify API

---

Why screen text for Claude?

New Claude models (from August 2026) embed a SynthID-Text-style watermark, but you cannot read it without Anthropic’s key. Teams still need a repeatable JSON label for documents they already hold.

Typical jobs:

1. Policy teams — log likely-Claude drafts while waiting for Anthropic’s detector API
2. Admissions integrity — screen essays and personal statements
3. Hiring take-homes — flag fully generated writeups
4. Research labeling — tag corpora as likely Claude vs human
5. Support QA — inspect long agent replies
6. SEO review — catch undifferentiated generated articles
7. Newsroom intake — label freelance copy
8. Education tools — show confidence, not a red-card verdict
9. Brand voice audits — find Claude-shaped pages
10. Legal extras — attach a classifier score next to produced files
11. Community moderation — sample long posts
12. Internal audits — sample exported docs for Claude-heavy writing

---

How to use Anthropic Claude Watermark Checker

1. Open the Actor in Apify Console
2. Paste Text / Texts, or upload Files
3. Keep Max characters per item at the default unless you know you need to send a longer sample
4. Click Start
5. Download the dataset or read it via the Apify API

Input example

{
  "text": "Four score and seven years ago our fathers brought forth on this continent, a new nation...",
  "texts": [
    "In today's rapidly evolving digital landscape, it is important to note that leveraging synergistic paradigms..."
  ],
  "files": ["https://example.com/memo.docx"],
  "maxItems": 5,
  "maxCharsPerItem": 24000
}

Field	Type	Default	Description
text	string	—	One pasted passage
texts	string[]	—	Extra pasted passages
files	string[]	—	Uploaded file URLs (PDF, DOCX, or text)
maxItems	integer	5	Cap on pasted samples + files this run
maxCharsPerItem	integer	24000	Max characters sent to the detector per item (hard max 100,000)

---

Output

Each dataset item is one check. The headline field is isAiGenerated. model is the detector family (synthid-text), not a specific chat model.

{
  "id": "a1b2c3d4e5f67890",
  "sourceType": "text",
  "sourceName": "pasted-text-1",
  "sourceUrl": null,
  "textPreview": "Four score and seven years ago...",
  "textLength": 312,
  "wordCount": 58,
  "truncated": false,
  "isAiGenerated": false,
  "watermarkDetected": null,
  "watermarkType": null,
  "matchesTargetProvider": false,
  "likelySource": "human",
  "likelihood": "low",
  "confidence": 0.64,
  "detectionMethod": "classifier",
  "rationale": "Well-known human speech; no access to Anthropic's SynthID-Text key.",
  "limitations": "Claude watermarks use Anthropic's key. This run is a writing-style classifier.",
  "inputTokens": 940,
  "outputTokens": 175,
  "billedTokenUnits": 1,
  "model": "synthid-text",
  "targetProvider": "anthropic-claude",
  "error": null,
  "checkedAt": "2026-08-26T12:00:00.000Z"
}

---

Pricing

Long files use more tokens than a tweet. This Actor passes that cost through with a margin.

Event	When	Price (FREE)
text-checked	Once per successful analysis	$0.005
input-1k-tokens	Per 1,000 input tokens of the sample actually sent (rounded up)	$0.001
Actor start	Per run (× memory GB)	platform default

Failed downloads are not billed.

Examples (excluding start fee):

• ~500-word paste (~800 tokens) → ≈ $0.006
• 24,000-character file sample → ≈ $0.012

Leave maxCharsPerItem at 24,000 unless you have a reason to send more.

---

Files and token safety

• Text is extracted locally. Images, audio, and zip archives are rejected.
• Files over 15 MB are rejected.
• Longer documents are sampled up to maxCharsPerItem.
• Short samples (under ~50 words) are weak evidence.

For a Google SynthID-Text check, use Google Gemini Watermark Checker. For ChatGPT images and audio (the media OpenAI actually watermarks), use OpenAI ChatGPT Watermark Checker.

---

Is it legal to check AI-generated text?

This Actor only analyzes content you provide. It does not log into Claude. Do not treat the score as legal proof of authorship. A Claude watermark answers “was Claude involved?”, not who the user was.

---

Tips

• Prefer longer passages — Anthropic notes watermark confidence rises with length
• Upload text-based PDFs, not scanned image-only pages
• Watch billedTokenUnits if a run costs more than you expected
• Start with maxItems 1–3 while you learn the output shape

---

Image Credit

Image credit: Anthropic