UK probes legality of frontier AI training amid startup boom

The UK has launched a formal inquiry into whether training frontier AI models on publicly available data violates existing privacy laws. The probe, first reported by Sifted, could reshape how startups and large firms source training data for AI systems, particularly around consent requirements.

The timing coincides with a surge of UK-based AI startups securing funding for agentic and enterprise-grade tools. AI Score, which raised $5.4 million earlier this month to expand its AI governance platform, and Tandem Health, which closed a £75 million Series B last week to build an AI-native clinic operating system, both rely on large-scale data ingestion. A ruling against current training practices could disrupt product roadmaps and increase compliance costs, particularly for startups lacking resources to navigate regulatory ambiguity.

Industry observers note the inquiry mirrors broader tensions, where legal challenges and evolving policies have forced companies to adjust data strategies. Unlike some regions with clear risk-based rules, the UK’s approach remains principles-driven, leaving startups to interpret how privacy laws apply to AI training. The investigation may clarify whether anonymization or opt-out mechanisms suffice—or if more stringent requirements become mandatory.

For investors, the uncertainty arrives as capital continues flowing into UK AI ventures. Early-stage funds backing startups like Interhuman AI may face heightened due diligence on portfolio companies’ data practices. The outcome could also influence AMD’s reported local AI platform, which aims to run AI agents on endpoints, potentially reducing reliance on cloud-based training data.

No timeline has been set for the review, but startups are already preparing for change. Magic AI, which raised £8 million last month for its AI-powered fitness mirror, said it is auditing its data sources to preempt potential shifts. How quickly the UK clarifies its stance—and whether it aligns with or diverges from other markets—will determine whether startups double down on local innovation or seek jurisdictions with clearer rules.

Sources: sifted.eu