SignSplit, a company founded quietly in 2024, launched Monday with $400 million in strategic seed funding from W Group at a $1 billion valuation, SiliconANGLE reported. The startup is building a system for people to attach verifiable consent and payment terms to the data, voices and likenesses that AI developers want to train on.

What Happened

W Group, the sole investor named, is a consortium of 11 fintech firms focused on blockchain technology. Its founder, Volodymyr Nosov, said the group plans to commit more capital and resources so SignSplit can roll out globally at speed, according to SiliconANGLE.

SignSplit's core idea is what it calls signed data. An individual digitally signs a dataset, a recording of their voice, their likeness or other creative work, and that signature records consent and the conditions under which the material may be used. The company's slogan, as SiliconANGLE summarized it, is to sign the data and split the value.

RelatedNvidia Plans to Invest in Inference Chip Startup d-Matrix, The Information Reports

The platform has several layers:

  • Data pools: contributors add a given kind of data to a shared pool and are paid based on how often that data is used.
  • Company pools: businesses describe the real-world data or skills they need and recruit contributors who accept their terms directly.
  • Research pools: universities and life sciences companies can find participants for one-time or ongoing studies.
  • Verification: organizations can check whether a dataset or likeness has been signed and what conditions are attached before using it.

CEO and co-founder Alessandro Monterosso framed the product around AI moving into the physical world, arguing that human knowledge is becoming central to how the technology develops and that contributors should be both protected and paid.

Why It Matters

Where training data comes from has become one of the most contested questions in AI. Developers have faced copyright lawsuits from authors, artists, music labels and news publishers over material collected from the web without permission, and several have since signed licensing deals with publishers and stock-image libraries. SignSplit is betting that consent and provenance can be turned into infrastructure, with a market for licensed human data large enough to support both payments to individuals and a billion-dollar company.

The verification layer may prove the most commercially relevant piece. Companies building models increasingly need to document where their data came from, both for litigation risk and for disclosure rules such as the European Union's AI Act, which requires general-purpose model providers to publish summaries of their training content. A registry showing that a voice sample or dataset was signed for a given use would address that need directly.

Open Questions

The size of the round is unusual for a seed financing, and it comes from a single consortium rather than a group of established venture firms. SiliconANGLE's report did not name any AI developers that have agreed to buy data through SignSplit, disclose pricing or say how payments to contributors will be calculated in practice. Nor did it explain how signatures are recorded or how the verification layer confirms them, details that will matter to companies relying on it for legal cover.

What to Watch

The test is adoption on both sides of the market. SignSplit needs enough contributors to make its pools useful and enough model builders willing to pay for signed data rather than rely on cheaper sources. Watch for the first named customers, the payout rates offered to contributors and how quickly W Group's promised additional capital arrives.