Another Way to Look at Policy

Every Horizon Europe call is a policy decision wearing a budget. The topics, the eligibility rules, the evaluation criteria, the weight given to a consortium from three member states over a company from one: each was argued somewhere in Brussels before the money was announced, and the argument is readable in the text for anyone who has been taught to read it. Most applicants have not. They meet the call as a form, treat the policy priorities as boilerplate, and lose points to the competitor who understood that the boilerplate was the brief.

I spent three years at the European Commission’s front line taking that call, more than a thousand enquiries a year, most of them from researchers and founders who were technically excellent and institutionally lost. Before that I ran the conferences of a European university network of close to a hundred institutions, and one subject came up at every one of them, whatever the agenda said: EU funding, how to read it, how to get it, how to report on it. The pattern on the phone and in the conference hall was the same. The question was about a deadline or a form; the answer was about why the programme existed. Once the why was in place, the how got easier.

That experience is the case for putting European innovation policy into higher education and lifelong learning as a subject in its own right rather than a line in a careers service handout. Students will meet EU rules in their first job, in one form or another, and the destinations data shows how few of them escape it. Three in four master’s and doctoral graduates of the University of Luxembourg take their first job in the country, in banking and finance, legal services, IT and education, every one of them a regulated or publicly funded field; 42.5 per cent of its doctoral graduates go to the private sector. At Heidelberg, between 82 and 89 per cent of graduates of the 2013 to 2015 cohorts were working outside academia within two years, while roughly a third had started a doctorate, which in Germany means a career lived on public research funding. Charles University in Prague sent 8,939 graduates into a market with 0.7 per cent graduate unemployment last year while launching 23 new Horizon Europe projects and running 17 ERC grants. Public service, private firms, research: three destinations, one requirement in common. Each of them will ask the graduate, within a year, to read a policy document and act on it.

The subject has two halves. The first is how Europe’s priorities become innovation and opportunity: how a political objective turns into a work programme, a work programme into a call, a call into a project, and a project into a product. The instruments carry the logic. Horizon Europe funds collaborative research across borders because the policy aim is a European research area, not a national one. The European Innovation Council funds single companies, with grants and equity, because the policy aim is deep-tech scale-ups that do not leave for California at Series B. Neither instrument makes sense to an applicant who has not been told what it is for, and both make perfect sense to one who has. That is the ground the programme European Innovation: From Policy to Funding covers: the priorities, the instruments, and the move from policy to implementation, where a funding opportunity becomes a practical project decision.

The second half is newer and less taught. The EU AI Act is read as a compliance text, and it is also, in its own words, a text about innovation. Article 57 obliges every member state to run an AI regulatory sandbox, a supervised environment in which providers and prospective providers develop, train, validate and test innovative systems under the eye of a competent authority; the Digital Omnibus moved the national deadline to 2 August 2027. Article 60 allows real-world testing of high-risk systems outside the sandbox, under conditions. Article 62 gives SMEs and start-ups priority access and reduced conformity fees. None of this is an exemption from the Act, and a sandbox is often misread as one. It is a controlled regulatory environment with defined conditions and supervision, and for a company building AI it changes the question from “are we compliant” to “how does regulation affect what we build, how we test it, where we launch it and how we scale it”.

That question has four stages. Build: not every AI activity sits in the same regulatory position, and the practical question is where the system stops being an experiment and becomes a product placed on the market or put into service. Test: the sandbox and real-world testing, which trade documentation and supervision for legal certainty. Scale: the Act harmonises the rules across the EU, so regulation becomes part of the market-entry conversation, and the intended purpose, the role under the Act and the standards that will matter are settled before expansion rather than after. Govern: a company does not stay in the same regulatory position as it grows; a system developed for internal use may become a product for customers, and governance has to develop alongside the product.

Europe is not the only jurisdiction writing these rules, and the comparison is where the subject becomes advanced. The United States and China have taken approaches to AI policy that differ from Europe’s and from each other, in strategic priorities, in the relationship between the state and the firms, and in what governance is for. Switzerland and the United Kingdom sit next to the EU market and have chosen their own paths, which matters to any Luxembourg organisation whose clients, suppliers or subsidiaries are across one of those borders. The programme AI Policy Global Development (Advanced) looks at several jurisdictions and asks what the differences mean for European organisations and decision makers.

A last point, about method. Technology is the subject of these programmes and also the instrument. A funding guide anyone can search, a tool that brings EU calls together in one place, a plain-language reference on the AI Act: I built each because the institutional knowledge they hold used to live in the heads of a few officials and reach the applicant one phone call at a time. A digital learning hub is the same idea at scale. Communication, learning and the exchange of ideas are the mechanism by which policy becomes practice, and a cohort that has argued over a real call and a real clause, in a format it can return to, leaves knowing something no summary gives it.

Both programmes are built for universities and for lifelong learning providers, from a single masterclass to a full module, and are adapted to the institution’s own cases.