Sep 26, 2024 14:00:00

More than 100 companies, including Amazon, Google, Microsoft and OpenAI, have signed the EU AI agreement as a 'voluntary commitment' before the AI law comes into full force in August 2026; Apple and Meta have not signed

In the EU,

the Artificial Intelligence (AI) Act was enacted in May 2024 with the aim of promoting the development and adoption of safe and reliable AI systems by private and public organizations throughout the European market, and a portion of it came into effect on August 1, 2024. However, since the full implementation of all provisions is scheduled for 2026, more than 100 companies, including OpenAI, have signed the 'EU AI Agreement' ahead of the full implementation.

Over a hundred companies sign EU AI Pact pledges
https://ec.europa.eu/commission/presscorner/detail/en/ip_24_4864

A Primer on the EU AI Act: What It Means for AI Providers and Deployers | OpenAI

https://openai.com/global-affairs/a-primer-on-the-eu-ai-act/

Meta will not immediately join EU's AI Pact ahead of new law | Reuters
https://www.reuters.com/technology/artificial-intelligence/meta-will-not-immediately-join-eus-ai-pact-ahead-new-law-2024-09-24/

Early sign-ups to EU's AI Pact include Amazon, Google, Microsoft, and OpenAI — but Apple and Meta are missing | TechCrunch
https://techcrunch.com/2024/09/25/early-sign-ups-to-eus-ai-pact-include-amazon-google-microsoft-and-openai-but-apple-and-meta-are-missing/

The AI law enacted in May 2024 sets out obligations for AI based on the potential risks and impact levels of AI, prohibiting policing based on predictions from profiling and the use of AI to classify people according to specific categories such as race, religion, and sexual orientation. In addition, an 'AI Office' directly under the European Commission will be established for appropriate enforcement, as well as a committee composed of independent experts, an AI committee including representatives of member states, and an advisory council to provide technical expertise to the AI committee and the European Commission.

Violations of the AI law can result in fines of up to 35 million euros (approximately 5.6 billion yen) or 7% of the violating company's annual worldwide turnover for the previous financial year.

EU Artificial Intelligence (AI) Law approved by member states aims to 'promote the development and deployment of safe and reliable AI systems' - GIGAZINE

The AI Act went into effect on August 1, 2024, but the rules will not be applied to all companies operating in the EU until August 2026, creating a gap in enforcement of compliance violations that the EU is trying to fill with the AI agreement.

Therefore, 115 companies including Amazon, Microsoft, Google, and OpenAI have voluntarily signed the EU AI Agreement, which states that 'we will begin to apply the principles of AI law prior to the full application of AI law.' The EU AI Agreement requires participating companies to commit to the following three things:

Adopt an AI governance strategy to drive AI adoption within your organization and work towards future compliance with AI laws.
Identifying AI systems likely to be classified as high risk under AI law
・Ensure ethical and responsible AI development by adhering to staff AI awareness and literacy

In addition to these core commitments, more than half of the signatory companies have made additional pledges, including ensuring human oversight, mitigating risks, and providing transparent labeling for certain types of AI-generated content, such as deepfakes.

According to the European Commission, the EU AI Agreement was drafted by the AI Office, the body that oversees AI law, and was filtered through feedback from 'relevant stakeholders' in the AI Agreements Network.

'We are proud to support the core priorities of the EU AI Agreement, which align with our mission to deliver safe, cutting-edge technology that benefits everyone,' said Sandro Gianella, head of policy and partnerships for Europe and the Middle East at OpenAI. A spokesperson for Snap said, 'The AI Act's goal of supporting the development of trustworthy AI is fundamentally aligned with our values and ongoing efforts.'

At the time of writing, companies such as Anthropic, NVIDIA, Spotify, Apple, and Meta have not signed the EU AI Agreement. The European Commission said, 'Until the AI law is fully applied, we are welcome to join the EU AI Agreement and commit to its provisions at any time.' A Meta spokesperson also said, 'We welcome harmonized EU rules and are focused on compliance activities under the AI law,' and did not deny the possibility of signing the EU AI Agreement in the future.