A recent mapping by Sifted identifies 37 European startups working on AI safety, a field that has jumped up the public agenda as more capable models and agents raise questions about safeguards. These companies are collectively building what the report calls a "trust layer" for AI, addressing risks that range from technical failures to existential threats.
The startups operate across several niches, including model evaluation, red-teaming, alignment research, and governance tools. Some focus on technical audits, while others provide compliance and risk-management platforms. The sector is still young, but it is attracting meaningful investment: funding rounds in the mapped companies range from early-stage seed deals to larger raises of tens of millions of euros.
The growth of this ecosystem reflects a broader shift in how businesses and regulators view AI risk. As AI adoption accelerates, the demand for independent safety and verification services is likely to increase, positioning these startups as critical infrastructure for the European AI economy.