AI tops €550M digital R&D agenda in Horizon Europe, leaked document shows

If you’re having a hard time keeping track of the new artificial intelligence and data laws in Europe, then you’re in luck: as part of its planning for Horizon Europe research grants, the Commission is considering harnessing AI to monitor compliance with those very laws, according to a draft document obtained by Science|Business.

Following the EU’s adoption in recent years of the General Data Privacy Regulation, the AI Act, the Open Data Directive and other laws, organisations, “face increasing challenges in maintaining compliance,” the Commission document says. “The complexity and volume of reporting obligations are growing, posing difficulties for both regulatory bodies to enforce laws and for entities trying to comply.”

One response could be to fund research consortia to explore ways to solve these and related problems, including an AI “auto-compliance” system. To get it started, the Commission may allocate €45 million under Horizon Europe, to be divided up into projects getting €7 million to €9 million apiece.

Leaked draft

The plans are in a draft Horizon Europe document dated May 2024. It covers digital research the Commission is considering funding, starting in 2025, and covers AI, quantum, photonics, chips, networking and related fields. In all, the proposed calls would add up to €550 million, as part of a larger “cluster” of industrial technologies in Horizon Europe.   

None of these plans are set in stone yet, but it’s expected many of these ideas will end up in public calls for grant applications starting next year. Under the normal Horizon planning process, the Commission’s draft plans are discussed with member states and expert committees for several months before adoption. This draft plan is dated 30 May 2024, and has been circulating privately among key member state officials and their contacts. An extra layer of review will come this winter as the newly installed College of Commissioners takes office.

Though preliminary, the draft puts some flesh on the bones of AI policy the Commission has been working on. So far, US-based organisations – Open AI, Google, Microsoft and others – have led in the nascent market for “generative” AI like ChatGPT; Commission documents cite a McKinsey report saying that market could eventually amount to €2.4 trillion to €4 trillion a year. Chinese companies are also climbing fast, but no EU companies have been making a mark in the field.

The EU response to date has been adopting new AI regulation and announcing several ambitious projects to speed up development of a home-grown AI industry. But in May, the European Court of Auditors dinged the Commission for underspending on AI by €600 million due partly to Horizon’s late start in 2021.

Some of the calls under consideration are big, some small. One would set up a previously promised “hub” to interconnect European AI developers, as part of the GenAI4EU initiative. That initiative, announced as part of a big AI policy push last year by Commission President Ursula von der Leyen, is supposed to develop novel uses of AI in 14 industrial sectors.

The hub is intended to interconnect everybody involved in that, carry out a “comprehensive landscape analysis of generative AI initiatives,” and “support the uptake” of the resulting GenAI4EU solutions in other parts of Horizon Europe. It’s also supposed to “support the European Commission’s AI Office in its function to promote an innovative ecosystem of trustworthy AI.” The budget: €3 million, a relatively modest sum by tech industry standards.

Cooperation with Washington

Though EU industry is competing with the US in AI, the Commission also envisions putting in €9.4 million to carry out its part of a deal with Washington to work together on “AI for Public Good” projects. It would include EU procurement, rather than grants, of €2.4 million for AI-based imaging of breast and prostate cancer, €2 million for AI-enabled flood and fire disaster management, €2 million to use AI to help reconstruct important buildings destroyed in disasters or war (it cites Ukraine as a possible user), and €3 million for AI to help manage renewable energy networks, initially for possible deployment in Lima, Peru.

Other AI-related projects mooted in the document include €85 million for industrial automation systems, another €30 million specifically for AI in the pharma and aerospace sectors, and €20 million for “soft robotics” systems that can perform multiple, changing tasks. There is also €5 million for cooperation with African countries in AI.

Other parts of the draft digital plan:

• In quantum, €32.5 million could step up development of a “quantum internet”, so multiple quantum systems can interconnect over distances as great as 500 kilometres. This work would be limited to EU members and some allies, under rules permitting the Commission to restrict access to projects it considers too sensitive to share. Another €10 million is to get quantum working with standard computers; it too would be restricted. But €8 million would be budgeted specifically for quantum cooperation with South Korea, which just joined Horizon Europe this year. Another €3 million is planned to help the Commission develop more chip cooperation with Japan, Korea, the US, Canada and other allies.
   
• A few calls are under consideration as part of the Commission’s “3C” – connected collaborative computing networks – to support various telecoms and computing-related technologies. It includes €80 million to pilot a way for European telecommunications companies to better create an “ubiquitous mesh of computing and communications resources,” from consumer device to cloud and edge computing. This one would also limit international involvement.
   
• A €25 million photonics call is under consideration for new sensor technologies for “multi-modal” systems involving optical, biomedical, chemical and other types of sensors for healthcare, transport, agriculture and other industries.