The law vs AI – How not to accidentally cross legal lines as a founder

It’s tricky to navigate the waters when law meets a technology that’s evolving faster than regulation can catch up. 

As the UK and EU introduce new legislation like the AI Act and the Data Use and Access Act, founders are facing a growing list of questions: 

• How do you stay compliant while staying competitive? 
• What happens when your AI makes a mistake? 
• Who owns the outputs when machines “create”?

In this discussion, SeedLegals legal experts Josie Tamburello, Ben Conry, Sam Hammond and Emma Baumann reveal the details every startup should know about building, deploying and investing in AI, without crossing legal lines. Watch below.

Key takeaways

Getting your AI legals right from the start

• Use clear T&Cs or customer contracts depending on your business model – self-serve vs enterprise.
• Avoid promising accuracy in outputs – instead, include strong disclaimers and limit your liability.
• Define ownership of training data, model outputs and feedback early to avoid future disputes.
• Back your contracts with in-product safeguards like usage limits or warnings to prevent misuse.

Navigating privacy and compliance

• Assume GDPR applies – AI and data privacy are deeply interconnected.
• Know whether you’re a data controller or processor, as each carries different legal duties.
• Conduct Data Protection Impact Assessments (DPIAs) for any high-risk or novel AI use.
• The UK’s new Data Use and Access Act now allows more automated decision-making – with safeguards.

Protecting your intellectual property

• Establish clear IP ownership with employees and contractors
• Only humans can be inventors under UK law – AI-assisted works still need human authorship.
• Audit and license your training data sources – scraped or third-party content can breach terms.
• Clearly set out input vs output data rights in your contracts to protect both you and your users.

Preparing for investment in the AI era

• The BVCA’s 2025 model docs now include AI-specific warranties – investors expect compliance clarity.
• Be transparent about how your AI is used and regulated – investors want to understand your risk profile.
• Keep documents organised and accessible – from IP agreements to GDPR policies.
• Structure deals for flexibility and control, so you can pivot fast in a fast-changing AI landscape.