Products
AI transparency is no longer voluntary. In Australia, California, the EU, and New Zealand it is now mandatory for companies to disclose how they train their AI systems and how data moves through them. Any company operating across borders must answer to all of these regimes at once. But a published disclosure is just a claim. The question that matters is whether the model and the data behind it actually match what was declared.
We provide an independent testing service which determines whether a company’s AI practice matches its stated policies—in the jurisdictions where it operates, as well as the ones in which it sells its products. Using the same detection technology we built to detect creative works inside training datasets and models, we provide detailed and transparent audit reports for regulators and rights organizations tasked with evaluating a provider’s claims, as well as for companies verifying their own compliance—or a vendor’s—before someone else does.
Check what a company has published against what its models and datasets actually contain.
Stated policies are promises. We test whether day-to-day practice keeps them.
General-purpose AI providers must publish a sufficiently detailed summary of training content—using the AI Office’s mandatory template—and maintain a copyright policy that respects rights reservations. In force since August 2025, with AI Office enforcement and fines from August 2026.
Text-and-data mining is only permitted where rights holders haven’t reserved their rights. A provider’s copyright policy is testable: either reserved works stay out of the training pipeline, or they don’t.
Since January 2026, developers of generative AI systems available in California must publicly document their training datasets—sources, copyrighted material, and personal information included—and keep the disclosure current as systems change.
From December 2026, organizations covered by the Australian Privacy Principles must disclose in their privacy policies the kinds of personal information used in substantially automated decisions—and the Privacy Commissioner’s guidance sets clear expectations for AI data governance today.
New Zealand’s privacy principles apply across the AI lifecycle, and from May 2026 the new IPP 3A requires agencies that collect personal information indirectly—the way training data is typically gathered—to tell people about it.
The rulebook is still growing, and we track it as it does. As with all our work: we provide independent technical findings, not legal advice—your counsel and compliance teams decide what to do with them.
The company or model in question, the jurisdictions that apply, and the disclosures and policies to test—whether you’re a regulator, a rights organization, or the company itself.
We review the published disclosures, search the underlying datasets, and probe the models—independently checking whether stated policy matches observed practice.
Findings mapped to each jurisdiction’s requirements, with reproducible evidence for every gap—and ongoing monitoring as models and obligations change.
Whether you’re checking a provider’s claims, a vendor’s pipeline, or your own compliance before regulators do—an independent audit starts with a conversation.
