EU Countries Approve New Act to Advance AI Development
On 2nd February, Germany and France withdrew their objection to allowing advanced AI development. The EU has proceeded to further the regulations for AI projects. On this account, various EU member nations have voted in favor of the finalization of the EU’s AI Act.
The Commissioner of Internal Market of the EU, Thierry Breton, has talked about the approval of the political agreement approved in December 2023. The bill in question got a unanimous approval last year with all 27 member states voting in favor. Breton noted in a recent X post that the EU’s AI Act is a historical decision and the first of its kind in the world.
The AI Act includes risk-management strategies for regulating AI protocols. This framework also provides the basis for state adoption of AI technology in sectors such as biometric surveillance.
Additionally, it also regulates independent projects such as ChatGPT and brings more transparency requirements for such projects before they are introduced to the mainstream markets. The notification on social media by the Belgian Presidency Council also noted that the compromise text of the AI bill has been finalized.
EU’s AI Act Gains COREPER Mandates
COREPER stands for Committee of the Permanent Representatives of Governments of Member States to the European Union. As per media reports, legislators have started the preparations to convert the AI Act framework into a final compromise text.
The process was finalized after receiving COREPER I approval last week. Legislators have continued to express concerns regarding deepfake technology that allows anyone to impersonate other people using a VR filter perfected by AI.
The popularity and public access of deepfakes have allowed members of the public to post fabricated messages on social media sites and muddled the line between truth and fiction. This issue has created the problem of difficulty in discerning between truth and lies for investigators, law enforcement agencies, and the general public.
The Executive Vice President of EU Digital Age Margrethe Vestager said on Friday that the agreement is a positive step towards consolidating the AI Act. She noted that based on the risk component of AI tech such as use in job screening or university admissions may put greater liability on the developers. She stated that the AI Act focuses on high-risk use cases.
European AI Act Enters the Next Stages of Approval
Once the legislators have converted the AI Act into a final compromise text, the next step is to forward the proposal to EU’s legislative committee. The committee vote will determine if the bill is ready to be added to the legislature on 13th February.
Once the Act is passed out of the legislative committee, it will be forwarded to the European Parliament for consideration in March or April. If the act is finalized it is expected to go into effect by 2026. Meanwhile, it is possible that specified portions of the legislative framework may become applicable with immediate effect.
The Commission of the EU has also approved plans to make an AI office. The office will supervise regulatory compliance and work in tandem with the foundational models that are thought to possess systemic lags.
The legislative framework has also introduced measures to support AI developers. This policy has considered projects such as promoting supercomputer programs that will serve as training modulators for new and existing generative AI models.
