The European Union’s AI Act, a significant regulation for artificial intelligence applications, has been officially published in the bloc’s Official Journal. The law will go into effect on August 1, with full applicability expected by mid-2026. However, there are various deadlines to be aware of between now and then as different provisions will start to apply.
Lawmakers in the EU reached a political agreement on the comprehensive rulebook for AI last December. The regulation imposes different obligations on AI developers based on the perceived risk of their applications. While most AI uses are considered low risk and will not be regulated, certain high-risk use cases such as biometric applications, AI in law enforcement, and critical infrastructure will face stricter requirements around data quality and anti-bias.
For developers of general-purpose AI models like OpenAI’s GPT, there are transparency requirements to follow. The most powerful AI models may also be required to conduct systemic risk assessments. However, there was lobbying from some AI industry players and a few Member States to lessen the obligations on these models, citing concerns about hindering Europe’s ability to compete with the US and China in AI development.
The implementation of the AI Act will be phased. Prohibited uses of AI will be enforced six months after the law takes effect, including bans on activities like social credit scoring and real-time remote biometrics by law enforcement without specific exceptions. Codes of practice for AI developers will be in place around nine months after the law is enacted, with the EU’s AI Office responsible for providing these guidelines. There are concerns about who will draft the codes, with worries that AI industry influence could shape the rules.
Another key deadline is set for 12 months after the law comes into force, focusing on transparency requirements for general-purpose AI models. Some high-risk AI systems have a more extended compliance deadline of up to 36 months, while others must comply within 24 months.
Overall, the EU’s AI Act aims to set a regulatory framework for AI applications, balancing innovation with risk management in the rapidly evolving field of artificial intelligence. With clear deadlines and specific requirements for different types of AI systems, developers will need to ensure compliance with the new regulations to operate within the EU market.
