Evidence you can check
CrawlQ keeps the record of every human approval — and lets you prove it was not changed.
Most AI tools forget. Context decays between prompts, people, and model versions, and no one can reconstruct why a decision was made. CrawlQ keeps your research, brand context and approved decisions in your workspace. And the record of a human approval isn’t just stored — it’s verifiable against a public log CrawlQ cannot alter, as the CrawlQ control below shows.
GraQle project · EUR-Lex Drift-Guard
The GraQle project’s public CI reads the regulation text weekly — and flags drift.
A public, scheduled CI workflow in the GraQle repository re-hashes every EUR-Lex page that GraQle’s compliance documentation cites, every week. If the regulator changes a page, an issue opens automatically. It covers that documentation, not your CrawlQ workspace.
CrawlQ Studio · Tamper-Evident Audit Records
Don’t take our word for it. Neither should your auditor.
When a human approves AI content in Studio, that decision is written to a public record we cannot edit or delete from — the same kind of transparency log that secures the world’s open-source software. If a single word of an approved decision were changed after the fact, it would no longer match the public record — and anyone holding your exported trail can run that check. So when your auditor asks “how do I know this trail wasn’t edited?”, the answer isn’t “trust CrawlQ’s database.” You hand them the export; they confirm it against a public log CrawlQ cannot alter. And privacy is built in: only a fingerprint of the record is ever made public — never your content, your reviewers’ names, or your workspace.
Why this matters now
Regulators everywhere are converging on the same demand.
The EU AI Act’s transparency obligations apply from 2 August 2026, with penalties reaching €15M or 3% of global turnover — and the EU is not alone. The US FTC, the UK, Australia and Canada are all moving the same way: a human must stand behind AI content, and the record of that decision must be trustworthy, not editable after the fact. The EU standard is the strictest, so meeting it puts you ahead of the rest. Most AI tools give you a log inside their own database and ask you to trust it. We put the fingerprint of your record in a public log CrawlQ cannot edit or delete from — so the integrity of your evidence doesn’t rest on trusting a vendor, including us.
How it works in practice: you hand your auditor the exported trail (from you, not from us), and the public log lets them confirm it hasn’t been altered since it was anchored. Technically minded? The exact reproduction steps — pull the public entry, match the fingerprint, check the signature — live on our engineering disclosure, with a worked example against our first live anchor.
GraQle SDK: compliance mode is a cryptographic latch
EU-AI-Act mode is an ed25519-signed, hash-chained, irreversible switch. Once enabled it cannot be silently disabled — a tamper attempt fails closed.
GraQle repository: 13 public EU AI Act article-mapping documents
Article-by-article mapping (Articles 4, 12, 13, 14, 15, 25, 43, 50 and supporting schemas), written to be quoted in your own compliance file. Read them.
GraQle SDK: non-claims discipline, enforced in code
A release-blocking invariant test scans every governance record and refuses to ship if a “compliant” or “certified” field appears. The vocabulary discipline your legal team wants is a test suite, not a style guide.
The GraQle SDK is open-core engineering you can audit: 150k+ lines of Python, 440 public test files (~7,700 test functions), public CI on Python 3.10–3.12, 140+ PyPI releases, and a reproducible MultiGov-30 benchmark harness — verify yourself. Full engineering disclosure → Meet GraQle, the SDK itself →
Three European patent applications pending on the underlying methods (patent-pending). Controls labelled GraQle describe that separate open-source SDK only; they are not features of your CrawlQ workspace. Not legal advice.
