Spain rejects AI self-regulation with new Plan IA360 accountability framework

Spain's aggressive stance signals a shift from voluntary safety pledges to hard legal liability for AI developers. By demanding financial accountability for algorithmic harms like IP theft and deepfakes, the government is placing the burden of proof on tech firms rather than users. This move challenges the 'slow down' narratives from leaders like Dario Amodei and Sam Altman, framing their safety warnings as potential fundraising tactics rather than policy solutions. For the streaming ecosystem, this creates a blueprint for how European regulators may soon enforce copyright protections against generative models. Watch for the passage of Spain's national AI bill in 2027 to see if these strict liability standards become permanent law.

Spain's Plan IA360 arrives amid a broader European regulatory momentum that extends well beyond Madrid. The Barcelona Supercomputing Center, which will host the AI gigafactory component of the plan, has already been designated as one of the EU's AI Factories under the European Commission's 2025 initiative to build sovereign compute infrastructure. In March 2026, the European Commission confirmed that Spain's MareNostrum 5 supercomputer would receive additional funding to scale AI training capacity under the AI Factories program, positioning the country as a compute hub for southern Europe. AESIA, Spain's AI supervision agency established in 2023, is expected to serve as the enforcement body for the Social Contract provisions, giving it a mandate comparable to national data protection authorities under GDPR.

The regulatory and licensing dimension of Plan IA360 intersects directly with the EU AI Act compliance timeline. The Act's provisions on general-purpose AI models became enforceable in August 2025, and the European AI Office published its first set of guidelines for GPAI model providers in July 2025, requiring transparency documentation and copyright compliance plans. Spain's insistence on financial liability for algorithmic outputs, including intellectual property theft, goes further than the EU baseline by shifting from documentation obligations to direct monetary penalties. Sánchez's framing of industry safety warnings as fundraising tactics also echoes criticism from European policymakers; in February 2026, European Commission Executive Vice President Henna Virkkunen stated that voluntary safety commitments from AI labs were insufficient without binding enforcement mechanisms, reinforcing the bloc's preference for hard law over self-regulation.

For the streaming and video industry, Plan IA360's intellectual property enforcement provisions carry direct implications for generative AI tools used in content production. The EU AI Act already requires GPAI providers to respect copyright opt-outs under the Copyright Directive's text and data mining exception, and the European Commission's July 2025 guidelines specified that model providers must demonstrate compliance with rights-holder reservations. Spain's additional liability layer means that a streaming platform using AI-generated content could face financial penalties if the underlying model was trained on copyrighted material without proper licensing. This creates a compliance calculus for video companies operating in Spain that mirrors the GDPR's extraterritorial reach: any service accessible to Spanish users may fall under the Social Contract's accountability framework once Spain's national AI bill passes in 2027.