Tech giants split on EU AI code as compliance deadline looms

The implementation of the General Purpose Persons Blog in the European Union revealed deep sections between major technology companies. Microsoft has indicated its intention to sign the European Union’s compliance framework for the European Union’s compliance framework while Meta categorically shared participation, describing the organizational guidelines that will suffocate innovation.

The head of Microsoft Brad Smith said Reuters On Friday, “I think we are likely to sign. We need to read documents.” Smith emphasized the approach of his cooperative company, saying: “Our goal is to find a way to be supportive, and at the same time, one of the things we welcome is direct participation by the Artificial Intelligence Office with the industry.”

On the other hand, Joel Kaplan, the chief global affairs official at Meta, announced on LinkedIn that “Meta will not sign it. The blog provides many legal uncertainties for moderate developers, as well as measures that exceed the scope of the AI law.”

Kaplan argued that “Europe is heading to the wrong path to artificial intelligence” and warned that the European Union symbol would “suffocate and spread the AI border models in Europe, and the European companies that look forward to building companies over them.”

The first adoption for admission

The response of the broken technology sector highlights the various strategies to manage European organizational compliance. Openai and Mistral fell on the symbol, where they put themselves as a first -time adoption of the volunteer framework.

Openai announced its commitment, saying: “The signing of the blog reflects our commitment to providing the capabilities of Amnesty International, which is capable and safe for Europeans to fully participate in the economic and societal benefits of the era of intelligence.”

Openai joins the European Union’s practice blog for artificial intelligence models for general purposes, the second signature of the leading AI after Mistral, according to industry monitors who follow voluntary obligations.

More than 40 of the largest companies in Europe signed a speech earlier this month, asking the European Commission to stop the implementation of the artificial intelligence law, including companies such as ASML Holding and Airbus, which called for a two -year delay.

Code requirements and schedule

The Practice Blog was published, on July 10, by the European Commission, and aims to provide legal certainty of companies that develop Amnesty International Public Purposes forms before compulsory implementation starting from August 2, 2025.

The volunteer tool has been developed by 13 independent experts, with inputs from more than 1,000 stakeholders, including models, small and medium enterprises, academics, artificial intelligence safety experts, law holders, and civil society organizations.

The European Union symbol puts artificial intelligence requirements in three areas. Transparency obligations require service providers to preserve the technical model documents and the data group, while compliance with copyrights imposes clear internal policies that define how to obtain training data and use them under copyright rules in the European Union.

For the most advanced models, safety and security obligations apply under the category, “GPAI with regular risks” (GPAISR), which cover the most advanced models, such as Openai’s O3, Claude 4 OPUS, and Gews Gemini 2.5 Pro.

The sites will have to publish summaries of the content used to train artificial intelligence models for general purposes and develop a policy to comply with the Copyright Law in the European Union. The framework requires companies to document training data sources, implement strong risk assessments, and to create governance frameworks to manage potential artificial intelligence system threats.

Implementation and penalties

Penalties on non -compliance are large, including up to 35 million euros or 7 % of an annual global rotation (larger than either of them). In particular, for service providers in GPAI models, EC may impose a fine of up to 15 million euros or 3 % of the annual world rotation around the world.

The committee indicated that if the adherence to the approved practice is adhered to, the Bureau of IQ and the national organizers will deal that as a simplified compliance path, with a focus on implementation on verifying the fulfillment of the blog’s obligations, instead of conducting audits for each Amnesty International system. This creates incentives for early adoption between companies that seek to obtain organizational capacity.

The European Union symbol is part of the Broadrai framework. Under the AI Law, the obligations for GPAI models, detailed in Articles 50-55, are implemented after twelve months of the law entered into force (August 2, 2025). GPAI models that were placed on the market before this date need to comply with the AI August 2, 2027.

The impact of industry and global monuments

Various responses indicate that technology companies adopt various strategies mainly to manage organizational relations in global markets. Microsoft’s cooperative position is sharply contrasting with the Meta confrontation approach, and it may determine precedents of how the main artificial intelligence developers interact with the international organization.

Despite the escalating opposition, the European Commission refused to delay. The European Union’s Internal Market Commissioner, Terry Bretton, insisted that this framework will continue as specified, saying that the artificial intelligence law is necessary for consumer safety and confidence in emerging technologies.

The current voluntary nature of the European Union symbol during the initial stages of companies provides opportunities to influence organizational development through participation. However, the mandatory enforcement that begins in August 2025 guarantees the final compliance regardless of the adoption of volunteer law.

For companies working in multiple judicial authorities, the European Union framework may affect the standards of global artificial intelligence governance. The framework is compatible with the developments of wider global artificial intelligence, including the G7 Hiroshima Ai process and various national artificial intelligence strategies, which may create European curricula as international standards.

We look forward

In the direct semester, the content of the blog will be reviewed by the European Union authorities: the European Commission and the member states evaluate the adequacy of the blog and are expected to formally support it, with a final decision planned by August 2, 2025.

The organizational framework creates great implications for developing artificial intelligence worldwide, as companies must balance innovation goals with compliance obligations in multiple judicial states. The company’s various responses to volunteer law predicted the potential challenges with the intervention of mandatory requirements.

See also: Mobility in the European Union Law, Amnesty International: The effects of UK’s business