Spain Introduces Strict AI Labelling Law to Combat Deepfakes
Spain's government has approved a new bill imposing hefty fines on companies that fail to properly label AI-generated content. The legislation aims to curb the spread of deepfakes videos, images, or audio manipulated or created by artificial intelligence but presented as real.
The bill aligns with the European Union’s AI Act, which enforces strict transparency rules on high-risk AI systems, according to Digital Transformation Minister Óscar López. He emphasized that while AI has the potential to enhance lives, it can also be used to spread misinformation and undermine democracy. Spain is among the first EU nations to implement these regulations, which are considered more rigorous than the largely voluntary approach in the United States.
Under the proposed law, failing to label AI-generated content correctly will be classified as a "serious offense," carrying penalties of up to €35 million ($38.2 million) or 7% of a company’s global annual revenue.
The bill also bans certain AI practices, including the use of subliminal techniques—imperceptible sounds and images—to manipulate vulnerable groups. López cited examples such as chatbots encouraging gambling addiction or AI-powered toys persuading children to engage in dangerous activities.
Additionally, the legislation prohibits organizations from using AI to classify people based on biometric data, assessing their behavior or traits to determine eligibility for benefits or predicting potential criminal activity. However, real-time biometric surveillance in public spaces will still be allowed for national security purposes.
A newly established AI supervisory agency, AESIA, will be responsible for enforcing the law. However, cases related to data privacy, crime, elections, credit ratings, insurance, or financial markets will be overseen by the relevant regulatory bodies.
The bill now awaits approval from Spain’s lower house of parliament before it can take effect.
