Herta, a Barcelona-based facial recognition and video analytics company, has successfully completed Spain’s artificial intelligence regulatory sandbox, the first such program in the European Union, putting its BioSurveillance system through a structured assessment against the EU AI Act. By its own account, the company emerges as the first European facial recognition specialist prepared to take on public-space projects in line with the AI Act, in the cases the regulation permits and under the safeguards it requires.
The sandbox is a controlled environment for helping high-risk AI systems meet the new rules. Run by Spain’s State Secretariat for Digitalisation and Artificial Intelligence under Royal Decree 817/2023, it ran from May 2025 to June 2026 and closed with an event in Madrid on June 11. Over that period, participants worked to align their high-risk AI systems with the AI Act, received legal and technical guidance, and helped shape official technical guidelines. Herta took part with BioSurveillance, its video surveillance system that uses facial recognition to identify people in crowded and dynamic settings, and earned an accreditation recognizing that work.
“Completing the European Union’s first Artificial Intelligence Sandbox is a strategic milestone for Herta. It demonstrates that it is possible to develop and deploy facial recognition technology from Europe with the highest legal, technical and ethical safeguards,” said Javier Rodríguez Saeta, CEO of Herta. Reaching that point meant building BioSurveillance to satisfy the AI Act’s demands for human oversight, risk management, traceability, technical documentation, cybersecurity, and the protection of fundamental rights.
Those standards are demanding. The AI Act sets a high bar for facial recognition in public places, banning real-time remote biometric identification in publicly accessible spaces in most circumstances, permitting only narrow and separately authorized law-enforcement exceptions, and classifying other biometric identification systems as high-risk uses subject to strict controls. With the compliance regime still taking shape, including the European Parliament’s vote to delay the AI Act’s high-risk deadlines to December 2027, Herta’s choice to test its system early reflects a compliance-first approach to a sensitive market.
Herta positions BioSurveillance for public security, critical infrastructure, transport, large-scale events, border control, and high-traffic venues, and describes the result as a milestone for the wider European biometrics sector. The company already supplies facial recognition for transport networks, including a deal to provide facial recognition technology for Indian Railways, and its sandbox work lands as European authorities weigh how such systems should operate in public, as seen in a Baden-Württemberg coalition plan for CCTV combining facial recognition and behaviour analysis.
By taking BioSurveillance through the European Union’s first AI sandbox, Herta has put its facial recognition platform on a documented footing for the AI Act’s high-risk requirements.
Sources: Source Security
–
By the ID Tech Editorial Team
