One standard covers you everywhere. We built for it.

Seven regimes. One convergent demand.

Regulators on four continents wrote different laws that ask for the same two artifacts: a documented human review point and an audit trail that can’t be quietly edited. Article 50 packages both at the strictest level. That’s why we call it the ceiling — and why “built for Article 50” is a statement about your whole stack, not your EU traffic. Only these two obligations are universal across all seven regimes — and both are shipped. That’s not a coincidence; it’s why we built these two first.

How it works — the real product, four steps

1. 1 · Enable it — deliberately

   Compliance mode is opt-in per workspace and owner-only, confirmed by typing ENABLE COMPLIANCE. You see exactly how many existing assets will enter the review queue before you commit. Turning it off later takes 30 days — our own integrity safeguard (not a legal mandate), because compliance you can switch off the moment it's inconvenient isn't compliance.

2. 2 · Every AI output stops at a human

   No AI-generated asset can be downloaded or published until a named person approves it. The reviewer sees exactly what will publish — full, untruncated content — with the model and prompt that produced it.

3. 3 · Approval = editorial responsibility, on the record

   The reviewer's name goes on the compliance record. The button says what it does: “Approve & commit to audit trail.” The confirmation asks the real question: “Take editorial responsibility?” It attests the reviewer saw these exact bytes — if the content changed since they loaded it, the approval is refused and they re-review.

4. 4 · Export evidence any time

   Every approval is committed to a tamper-evident, Merkle-chained audit trail with a per-asset inclusion proof (a short cryptographic receipt tying that one asset to the chain) — what model made it, what it was built from, a content fingerprint, and who approved it. One click, yours to hand to counsel, a client, or an auditor.

What we don’t cover — yes, really

Trust is built by naming gaps before you find them. Here are ours, dated 15 July 2026:

• C2PA machine-readable content marking

  Planned, not shipped. Today our disclosure is human-readable and in-platform; embedded content credentials come later. Notably, the EU is the only one of the seven regimes that demands machine-readable marking.

• “Independently verifiable without contacting CrawlQ”

  Not yet. The audit chain is tamper-evident and exportable from your own workspace today; public transparency-log anchoring that lets a third party verify with zero trust in us is the next layer of our roadmap. We'll make that claim when it ships, not before.

• US state AI laws, CASL, and the Act's high-risk regime

  Not specifically covered. The growing US state patchwork (Colorado, Texas and others), Canada's CASL email-consent law, and the Act's Annex III high-risk regime (a separate, heavier rulebook for high-risk AI systems) are outside a content platform's scope. If you operate high-risk AI systems you need more than us.

• No tool makes you “compliant”

  Compliance is your legal state, assessed with your counsel. We provide the working mechanism and the evidence. If a vendor tells you otherwise, ask them for their gaps list.

Frequently asked questions