Brussels – On artificial intelligence (AI), the European Commission gets into the game early: the Brussels-proposed AI Pact has gathered over a hundred signatures. The signatories have pledged right away to implement and enforce the requirements of the AI Act, Europe’s first legal framework on artificial intelligence.
Implementing the AI Act measures, which on average will be applicable after two years, will require a challenging transition. The Commission has invited companies to express interest since November 2023 to support future changes. The result is voluntary industry commitment to anticipate the Artificial Intelligence Act and begin implementing its requirements before the legal deadline.
The pact on Artificial Intelligence has two fundamental pillars: collaboration and compliance. For the former, the goal is to disseminate companies’ know-how, with the acquisition of best practices from other participants in this preparatory phase. Another proposal is to provide an online platform covering the AI implementation process to foster exchange among companies.
The second pillar aims to outline a framework for implementing the standards. The assumption is that companies proactively disseminate implemented practices with other companies to facilitate alignment toward transparency goals and high-risk requirements, particularly for developers and resellers of artificial intelligence systems. Concrete commitments are mentioned, with ongoing or planned actions and associated timetables.
The goal is to build a solid foundation based on trust in AI technologies. Companies are asked to commit to creating a governance strategy, mapping “high-risk” systems, and disseminating AI knowledge among staff. In addition to these key objectives, over half of the signatories have committed to additional commitments, including ensuring human oversight, mitigating risks, and transparently labelling certain types of AI-generated content.
Among the participants, some giants, such as Meta, Apple, or Mistral, are missing. This does not bode well for the effectiveness of European standards.
From the Berlaymont, there is a firm answer: participation in the pact is voluntary and important for planning work that will have to be done in the future anyway. Signatories number more than a hundred, with Google and Microsoft among them (which, in the list of big companies, have their importance) but also some of the leading developers of artificial intelligence, such as OpenAI (creator of the well-known ChatGpt). “Artificial intelligence offers many benefits to all of us, but it also carries risks,” commented today Thomas Regnier, Commission spokesman, “we encourage all companies to join this initiative.”
In any case, companies will have to comply with the EU framework rules on artificial intelligence when they become effective. For those who have moved ahead by participating in the pact, the road may be smoother.
English version by the Translation Service of Withub
