Early participants in the EU’s AI Pact include Amazon, Google, Microsoft, and OpenAI, while Apple and Meta are notably absent.

The European Commission has announced the first 100-plus signatories to the AI Pact, an initiative aimed at encouraging companies to publish “voluntary pledges” regarding their approach to artificial intelligence.
Early participants in the EU’s AI Pact include Amazon, Google, Microsoft, and OpenAI, while Apple and Meta are notably absent.

The European Commission has announced the first 100-plus signatories to the AI Pact, an initiative aimed at encouraging companies to publish “voluntary pledges” regarding their approach to artificial intelligence.

While the EU’s legally binding, risk-based AI rulebook (the AI Act) came into effect last month, it will take several years for all compliance deadlines to be implemented. This gap presents an opportunity for the EU to address non-compliance through the AI Pact.

The initiative aims to enhance engagement and foster commitments, encouraging companies to proactively meet the law’s requirements before the deadlines. The Pact also emphasizes information sharing among signatories, enabling them to support one another in adapting to the incoming obligations of the EU's AI rulebook and to collaboratively develop best practices.

Signatories are expected to commit to three “core actions” at a minimum:

1. **Adopting an AI governance strategy** to promote the integration of AI within their organization and work towards future compliance with the AI Act.
2. **Identifying and mapping AI systems** likely to be classified as high-risk under the AI Act.
3. **Promoting AI awareness and literacy** among staff, ensuring ethical and responsible AI development.

In addition, a lengthy list of potential pledges (available in PDF form) was created by the AI Office, the body overseeing the AI Act, and refined based on feedback from relevant stakeholders within the AI Pact network. This list allows signatories to select commitments that align with their goals.

Examples of these pledges include commitments to “design AI systems intended to directly interact with individuals so that those individuals are informed, as appropriate, that they are interacting with an AI system,” and to “clearly and distinguishably label AI-generated content, including images, audio, or video constituting deep fakes.”

This extensive list may foster competition among signatories, encouraging them to enhance their AI safety practices.

### A Pact to Accelerate AI Act Compliance

The AI Pact initiative was unveiled in May 2023 by former internal market commissioner Thierry Breton, with Google agreeing to assist regulators in developing the initiative at that time. Now, over a year later, the EU has amassed many more signatories, although several notable companies are missing from the list.

For instance, Apple and Meta are not included. Meta informed Reuters on Tuesday that it would not join the effort immediately, opting instead to concentrate on its compliance work related to the AI Act itself.

The penalties for non-compliance with the EU’s legally binding AI rulebook are severe: fines can reach up to 7% of global annual revenue for violations involving prohibited uses of AI, up to 3% for non-compliance with other obligations under the AI Act, and up to 1.5% for providing incorrect information to regulators.

This could explain Meta's reluctance to join the Pact, as any misstep regarding the actual AI regulations could lead to substantial fines. Additionally, publicly abandoning pledges could invite criticism.

French large language model company Mistral is also absent from the list, which is not surprising given its strong opposition to the AI Act.

On the other hand, Germany’s Aleph Alpha, another large language model maker, has signed the Pact but recently announced a shift towards providing B2B support for generative AI tools. This change in business model may also influence its policy priorities.

Other companies that have signed the Pact include Amazon, Microsoft, OpenAI, Palantir, Samsung, SAP, Salesforce, Snap, Airbus, Porsche, Lenovo, and Qualcomm.

Notably absent from the list are Anthropic, Nvidia, and Spotify—especially the first two, given their importance in AI development.

Spotify's absence stands out as it recently signed an open letter organized by Meta, advocating against regulations that could hinder the advancement of generative AI.

The EU’s full list of early AI Pact sign-ups can be found here.

The signatories represent a diverse array of companies, including major European telecoms, consulting firms, software providers, banking and payment firms, multinationals, SMEs, and consumer-facing platforms. However, with over 100 names on the list, it still represents just the beginning of a much larger movement as generative AI technologies rapidly evolve.

Since these are purely voluntary pledges, signing the AI Pact may primarily serve as a strategy for enhancing reputational standing. Signatories are invited to report on their progress 12 months after publishing their chosen pledges, providing an opportunity for further publicity.

Blog
|
2024-09-30 23:06:57