EU and US set to announce joint working on AI safety, standards & R&D

The European Union and the United States plan to announce AI cooperation on Friday at a meeting of the EU-US Trade and Technology Council (TTC), according to a senior EU official. the Commission which gave journalists an overview of the context before the Confab.

The ambient music highlights growing cooperation between lawmakers on both sides of the Atlantic when it comes to strategizing to respond to the challenges and opportunities posed by powerful AI technologies – despite this which remains a very skewed business picture in which US giants like OpenAI continue to dominate cutting-edge AI developments.

The TTC was created a few years ago, after Trump’s election, to provide a forum where European and American lawmakers could meet to discuss transatlantic cooperation on trade and technology policy issues. Friday’s meeting, the sixth since the forum began operating in 2021, will be the last before elections in the two regions. The prospect of a second Trump presidency derailing future EU-US cooperation may well focus lawmakers’ minds on maximizing opportunities for joint work now.

“There will definitely be an announcement at the TTC regarding the AI ​​Office and the (US) AI Security Institute,” the senior Commission official said, referring to an EU watchdog that is being implemented as part of the next European AI project. Act, a comprehensive risk-based framework for regulating AI applications that will begin applying across the bloc later this year.

This element of the upcoming deal – apparently intended to focus on AI security or surveillance – is envisioned as a “collaboration or dialogue” between the EU’s respective AI watchdogs and of the United States, aimed at strengthening the implementation of regulatory powers in AI, according to the official. .

A second area of ​​focus for the expected EU-US deal on AI will be around standardization, they said. This will take the form of joint work aimed at developing standards that can support developments by establishing an “AI roadmap”.

The EU-US partnership will also have a third element, called “AI for the public good”. According to the Commission, this is joint work aimed at encouraging research activities, but with a focus on the implementation of AI technologies in developing countries and the Global South.

The official suggested that there is a shared perspective that AI technologies will be able to bring “highly quantifiable” benefits to developing regions – in areas such as health, agriculture and energy. This should therefore also be an area of ​​focus for transatlantic collaboration to drive AI adoption in the near term.

“AI” represents aligned interests?

AI is no longer considered a trade issue by the US, as the EU claims. “Thanks to the TTC, we were able to explain our policy and also show the Americans that in fact we are pursuing the same objectives,” suggested the Commission official. “Through the AI ​​Act and Decree (focused on AI safety and security), which aims to mitigate the risks of AI technologies while supporting their adoption in our economies.”

Earlier this week, the United States and the United Kingdom signed an AI security partnership agreement. Although the collaboration between the EU and the United States appears to have a broader scope, since it is expected to cover not only common objectives in security and standardization, but also aims to align efforts aimed at fostering adoption of AI in a number of third countries through joint support for “public good” research. .

The Commission representative highlighted other areas of collaboration on emerging technologies, including standardization work in the area of ​​electronic identity (for which the EU has been developing an electronic identification proposal for several years) which, according to them, would also be announced on Friday. “E-identity is a very strong area of ​​cooperation with a lot of potential,” they said, asserting that the United States is interested in “new and broad business opportunities” that the e-identity wallet will open up. the EU.

The official also suggested that there is growing agreement between the EU and US on how to manage the power of platforms – another area where the EU has targeted legislation in recent years. “We see a lot of commonality (between EU laws like the DMA, aka Digital Markets Act) with recent antitrust cases that are also being launched in the United States,” the official said, adding: “I think that in many of these areas there is no doubt that there is a win-win opportunity.

The US-UK memorandum of understanding on AI, signed Monday in Washington by US Commerce Secretary Gina Raimondo and UK Secretary of State for Technology Michelle Donelan, indicates that the two countries will aim to accelerate collaboration on a range of AI security issues, including in the area of ​​national security as well as broader societal AI security concerns.

The US-UK agreement provides for at least one joint testing exercise on a publicly available AI model, the UK Department of Science, Innovation and Technology (DSIT) said in a press release. He also suggested that there could be personnel exchanges between the two countries’ respective AI security institutes to collaborate on sharing expertise.

Wider information sharing is envisaged under the US-UK agreement – ​​on “capabilities and risks” associated with AI models and systems, and on “technical research fundamental on the safety and security of AI”. “This will help underpin a common approach to AI safety testing, allowing researchers on both sides of the Atlantic – and around the world – to come together around a common scientific basis,” continues the DSIT PR .

Last summer, ahead of hosting a global AI summit, the UK government said it had secured a commitment from US AI giants Anthropic, DeepMind and OpenAI to provide “early or priority access” to their models AI to support evaluation and security research. He also announced plans to spend £100 million on an AI security task force which he said would focus on so-called foundational or frontier AI models.

Meanwhile, at the UK AI Summit last November – hot on the heels of the US AI Executive Order – Raimondo announced the creation of a US AI Security Institute to be housed within its department, under the National Institute of Standards and Technology, which she said would aim to work closely with other AI security groups created by other governments.

Neither the US nor the UK have yet proposed comprehensive AI safety legislation – with the EU remaining ahead of the pack when it comes to legislating AI safety. But more cross-border collaboration seems obvious.