Consent infrastructure for AI training data.
Sourcemark makes creator choice visible, licensing routes clear and dataset decisions more auditable — before ingestion, not after.
Why Sourcemark exists
Creative work and written content are being used in AI training datasets at scale. Consent is fragmented. There is no consistent way for creators to signal whether their work is available for training, and no consistent way for AI companies to check.
The result is avoidable ambiguity. Creators are often silent by default, AI companies lack a consistent way to check, and questions surface after training — when the cost to both sides is higher.
Sourcemark shifts consent upstream. It gives creators a structured way to declare their AI training choice, and gives AI companies a machine-readable registry they can query before ingestion.
It does not replace regulation, licensing schemes or enforcement. It fills a gap they do not currently fill well: making consent visible earlier, when it can still be checked and acted on.
What we believe
Consent should be visible before training, not disputed after.
Sourcemark is built to make creator declarations usable earlier in the process, when they can still inform decisions.
Infrastructure works best when it serves both sides.
Creators need a clear way to declare their position. AI companies need a clear way to check it. The same record should support both.
Neutrality matters.
Sourcemark is not a campaign, a licensing marketplace or an enforcement body. It is infrastructure designed to make declared positions clearer and easier to act on.
Declaring intent should be simple and accessible.
The value of a consent registry depends on participation. We want it to be easy for creators to put their position on record.
We are honest about scope.
Sourcemark records declarations. It does not verify ownership, enforce consent or guarantee outcomes.
The team
Lyndsey Salah
Co-founder & CEO
More than 25 years in commercial stock photography, image licensing and visual content production. Built global lifestyle collections and worked across rights management at scale. Her recent work on volumetric datasets for generative AI exposed the gap that Sourcemark is built to close.
Nabil Salah
Co-founder & CTO
Bridges design and engineering across 20+ years — from the Electronic Telegraph to e-commerce platforms for luxury brands. Founded London studio Eat Sleep Breathe in 2002. At Sourcemark he architects the fingerprinting, registry and verification infrastructure.
Advisory board
Donia Dacey
Brand & Communications
More than 20 years in creative strategy and brand communications, with senior roles at Publicis Media, Snap Inc, TBWA\Media Arts Lab (Apple), News UK and BBC Worldwide. Leads brand positioning and go-to-market across both creator and AI company audiences.
Christiaan Haig
Talent & Organisational Design
Founder of EddisonFields, a London-based executive search and human capital advisory firm. Deep expertise in C-suite and board-level recruitment across financial services, infrastructure and technology. Leads talent strategy and organisational design as Sourcemark scales.
How Sourcemark fits
Sourcemark is designed to complement the wider infrastructure emerging around provenance, licensing and AI transparency.
Its role is narrower and practical: recording creator declarations in a form that can be checked before training.
Where provenance standards such as C2PA are already present, Sourcemark is designed to work alongside them.