Get in Touch
AI content toolsfrom Opace

Free browser checker · Evidence-led

Opace AI Content Verification, Integrity & Watermark Checker Tools

Inspect writing signals, protect exact facts and record which methods did not run. Every check but one runs on this page and sends nothing. The AI model scores on our EU server by default, or in your browser if you prefer, and no check turns a pattern prompt into proof of authorship.

Evidence sheetBrowser-only
SourceExact text and protected factsHeld in memory for this run
SignalsNamed checks with limitsNo universal authorship score
Watermark scanKnown keysReal SynthID mathematics; Anthropic production keys stay private
Evidence summary

Evidence stays attached to the method

See what ran, what changed and what remains unavailable.

The review keeps protected facts beside each proposed treatment and records unsupported methods without turning them into a score. The interface image shows the same evidence-led workflow used across the browser, WordPress, Chrome and Astro surfaces.

Opace AI Content Integrity evidence workflow with a protected rewrite review panel
Exact checks, protected facts and unsupported methods remain visible throughout the review.

02 / How the checker works

Check the text, protect facts, read the limits.

Each result keeps exact checks apart from editorial prompts and unsupported methods, so a missing detector can never look like a pass.

Inspect locally

Run exact character and lookalike checks, 116 named writing-signal rules across 113 categories including sentence rhythm, cadence and chat-export formatting, chatbot artefact traces attributed to the model families that leave them, and protected-content extraction, all in the browser. On a 5,558-document long-form corpus these rules flagged 45.1% of AI writing and 24.8% of human writing, so they are editing feedback and never an authorship verdict.

Keep methods separate

Every method has its own status, evidence and limitation. Uploaded images and PDF files get a local Content Credentials (C2PA) read; the beta AI model scores on our EU server by default and can be moved into your browser on explicit consent, with its measured accuracy always shown for the route that ran; unsupported and not-run checks stay visible.

Score the watermark keys

Every assessment runs the published SynthID-Text detection mathematics against three public demo keys in your browser and shows the per-key result. Anthropic production keys are private, so that watermark is reported as not assessed rather than guessed. Then export a hash-only receipt that records the run without silently retaining the draft.

03 / Where your text goes

Your text is never stored, and the checker tells you which route ran.

Every check runs in your browser except the AI model. The hidden-character, lookalike, protected-fact, writing and watermark checks send nothing. The AI model scores on Opace’s own server in the EU by default, in Belgium, over HTTPS: the whole document goes in one request, is scored in memory, then discarded. Your text is not stored or logged, and the request leaves no per-visitor record. Each result prints how many words were sent and repeats what the server says it did with them, and one click switches the model to run entirely in your browser instead. Either way, draft text, file names, hashes, findings, protected spans and receipt IDs never enter analytics, URLs or browser storage; the standard site analytics event the form fires carries no text, file or result content.

04 / Frequently asked questions

What the checker checks, and what it cannot claim.

What does content integrity mean here?

It means keeping exact text hygiene, factual anchors, method status and limitations visible while a draft is reviewed. It does not mean assigning a universal AI probability or claiming who wrote the text.

Which tools are available now?

The AI Content Integrity Checker is available on this site. The other four are local candidates and none has been submitted or published. Chrome 1.0.0 and the 0.1.0 Astro, command-line and loopback-service set passed their declared checks. The WordPress plugin is at 1.0.6, and its gate evidence was recorded against the earlier 1.0.4 build: rebuilding changed the plugin’s bytes, so Plugin Check, readme validation, the package inventory and the lifecycle, REST, editor, responsive, focus and axe checks have to be re-run against 1.0.6 before it can be submitted. Downloads and directory listings will appear only after final release approval.

Is the suite free?

The available browser tools are free. In the Content Integrity Checker every check runs in your browser except the AI model, which scores on Opace’s own EU server by default and can be switched to run in your browser instead; that page states which route is in use and what was sent.

What does the browser checker store?

No draft is kept. The checker holds the draft in browser memory for the current run, and on the default AI model route the whole document is sent to Opace’s EU server in one request, scored in memory and then discarded; your text is not stored or logged, the request leaves no per-visitor record, and each result prints how many words were sent. Our hosting platform keeps operational logs for 30 days that contain no part of your draft. Draft text, filenames, content hashes, findings, protected spans and receipt identifiers are never placed in the page URL, cookies, local storage or site analytics.

Where should I report a problem?

Use the Opace contact page with the product version, browser or platform version, expected behaviour and the smallest safe reproduction. Remove draft text, authentication tokens, personal data and customer information from screenshots or logs. Contact Opace.

Opace AI Content Integrity

Check the evidence. Keep the limits visible.

Use the free browser checker, then read the methodology before drawing a conclusion from any result.