European Parliament approves draft negotiating mandate for Artificial Intelligence Act

On the 14th of June, the European Parliament ratified the draft mandate for negotiations pertaining to the “Artificial Intelligence Act”. This constitutes the European Union’s preparation for regulations targeting artificial intelligence applications like ChatGPT, aiming to ensure artificial intelligence systems (AI) are subject to oversight, deciding on classifications for AI applications, and determining which activities are deemed impermissible.

Simulation of a screen of CCTV cameras with facial recognition

The EU Commission proposed the draft mandate for the “Artificial Intelligence Act” negotiations in April 2021. It stringently forbids AI systems that pose an intolerable risk to human safety, including deliberately manipulative technologies, systems exploiting human frailties, or those conducting assessments based on behavior, social status, and personal characteristics. It encompasses issues related to safety, transparency, privacy, and human oversight, providing a definitive list of AI use cases subject to direct prohibition. Furthermore, it aspires to address societal, environmental, and discriminatory matters via new regulations.

In addition to the prohibitions, the act stipulates certain requirements for developers, ensuring newly developed AI systems adhere to new regulations, such as the prerequisite of registering foundational models before their release into the market. Generative AI applications like ChatGPT must disclose that their content is AI-generated, and precautionary measures must be implemented to prevent the generation of illicit material.

In the European Parliament’s vote on the draft mandate for negotiations of the “Artificial Intelligence Act”, there were 499 votes in favor, 28 against, and an additional 93 abstentions.