Category: AI Careers

This blog post summarises the paper “Visibility into AI Agents,” by Alan Chan and a multi-institution team of co-authors. The paper is forthcoming at FAccT 2024.

GovAI research blog posts represent the views of their authors, rather than the views of the organisation.‍

‍

Introduction

Researchers are, increasingly, working to develop AI agents: AI systems that can pursue real-world goals with minimal supervision. 

Arguably, rudimentary AI agents already exist. For example, ChatGPT is no longer a simple question-and-answer chatbot. It can now perform tasks such as searching for information on the internet, writing and running computer code, and creating calendar events. 

Future agents may be capable of performing much longer and more complex sequences of tasks by themselves. They may eventually be assigned many important, open-ended responsibilities.

In part because they need less human steering and oversight, advanced AI agents could pose novel risks. One foundational concern is that people will not have enough visibility into their behaviour. For example, users of AI agents may not always know if their agents are misbehaving.

Therefore, we should make sure that key stakeholders will have enough visibility into AI agents. Obtaining visibility will mean thoughtfully collecting, processing, and sharing information about what agents are doing.

This post explores three particular measures that deployers — the companies that run AI agents for users — can take to create helpful forms of visibility. These measures are:

• Agent identifiers that allow third parties to know when they are interacting with an AI agent
• Real-time monitoring that allows deployers to notice agent misbehaviour, including misbehaviour that a user might not have noticed on their own
• Activity logs that allow deployers or users to review an agent’s past behaviour, to notice and better understand problems

Some of these measures are already in use for certain AI systems. In these cases, we discuss how to adapt them for agents.

More work is needed to understand how and when to implement these measures, since they also raise significant concerns around privacy and the abuse of power. Still, if implemented well, visibility measures could improve our understanding of AI agents, help reduce accidents and misuse, and generate data to inform governance decisions.

‍

AI agents and their risks

AI agents are AI systems that can pursue goals in the world under limited supervision. Current AI agents are still fairly rudimentary: they can only carry out certain short and simple tasks, such as creating calendar events and doing basic web searches. They struggle to perform many specific tasks, chain tasks together, and plan their actions.¹

However, researchers are working to create AI agents that can perform far longer and more complex sequences of tasks with minimal human supervision.² An advanced AI agent might operate more like a high-autonomy employee — who is given a high-level goal, pursues it with minimal oversight, and only occasionally reports back — and less like a simple tool. It could take on a range of personal, commercial, or even governmental responsibilities from people. These kinds of advanced AI agents could be very useful. 

At the same, AI agents could also pose significant risks. For example, AI agents may pose greater misuse risks than more tool-like AI systems, because they have broader capabilities and rely less on the skill of human users. AI agents may also have a greater potential to cause harm if they malfunction or suffer from controllability issues.³ New risks could also emerge from interactions between different AI agents, similar to the kinds of risks that we have already seen emerge from interactions between high-frequency trading bots.

Any risks from AI agents will be exacerbated because — by default — people will not have as much visibility into their behaviour: agents would accomplish tasks under limited supervision. Someone using an AI agent could therefore have only limited knowledge and understanding of what their agent is doing. Similarly, when an AI agent interacts with a third party, this third party might not even know they are interacting with an AI agent. In such cases, risks created by AI agents may remain unnoticed or poorly understood for long periods of time.

Hence, it will be especially important to ensure that key stakeholders have enough visibility into the behaviour of AI agents. This information could directly help deployers and users avoid accidents and instances of misuse. It could also offer insight into emerging risks and thereby inform governance decisions and safety research.⁴

‍

Obtaining visibility into the use of agents

We consider three ways to obtain visibility into the use of agents: agent identifiers, real-time monitoring, and activity logs. These measures could provide governments and civil society the information necessary to plan for and respond to the deployment of more advanced agents.

We focus on how deployers—companies like OpenAI and Anthropic that run agents for users—could implement visibility measures. Deployers are a useful intermediary because they tend to have access to all of an agent’s inputs and outputs, which facilitates implementation of visibility measures.⁵

‍

‍

First, an agent identifier is an indicator that would allow a person or software program to identify when they are interacting with an agent. Examples of agent identifiers include automated chatbot disclosures or watermarks, but many other kinds of agent identifiers could be created. For instance, when an agent performs a financial transaction with a bank, the deployer could send a unique ID associated with the agent to the bank. Additional information about the agent could be attached to the identifier as well, such as whether the agent has satisfied certain security standards.

The potential usefulness of AI agents means that they may be involved in many consequential interactions. Since agents would carry out activities with little to no human supervision, parties to an interaction would have the option to change how they interact with the agent. For instance, parties could refuse such interaction, or may take additional precautions to check the agent’s outputs. As such, agent identifiers could help participants in an interaction decide when and how to interact with AI agents. Widespread use of agents across a variety of consequential tasks may also be of interest to regulators. Aggregated information about agent identifiers could help measure the overall level of AI agent activity across various domains.

‍

‍

Second, real-time monitoring would involve oversight of agent activity in real time, whether by humans or software. Just as social media companies filter or flag abusive content, deployers could do the same for agent behaviours. Indeed, since agents may take consequential actions without users knowing, real-time monitoring may be especially important for ensuring that users approve the actions.

Some deployers already monitor model use for abuse, but monitoring protocols will need to be adapted for the wider variety of actions that agents could perform. For example, agents could carry out a variety of financial transactions on their users’ behalf. To reduce the risk of harmful or unintentional actions, financial transactions above a certain dollar amount could be flagged for user review. Moreover, monitoring protocols that focus upon a single system may be insufficient. If groups of agents can cause harmful feedback loops or other sources of instability, monitoring would need to account for their collective activities.

One limitation of real-time monitoring is that it is most effective when a human overseer or other computer program can swiftly assess a flagged behaviour and determine an appropriate course of action. Consequently, monitoring may be most useful in cases where the flagged behaviour clearly violates established policies or guidelines.

‍

‍

Third, activity logs could record some of an agent’s inputs and outputs for further analysis. Logs are helpful for understanding the risks of agents because harmful behaviour may only be recognisable as such long after its occurrence.⁶ Users or deployers reviewing logs may notice that an agent has taken a consequential action—which real-time monitoring might not have caught—without approval. As such, logs could inform real-time monitoring, help investigate incidents, and track long-run impacts of agent usage. 

Although logging is a common practice for software systems, agent-specific logging practices could be especially useful. Logs dedicated to certain parts of an agent could illuminate factors that affect an agent’s impact in the real-world. For example, memory or reasoning logs could help us understand why an agent malfunctioned in the pursuit of a complex, long-horizon goal. Logs about an agent’s use of tools or services could help us identify the impacts of an agent on third parties. It may also be important to analyse the logs of multiple agents so as to understand risks that come about from their interactions. 

Overall, visibility measures like agent identifiers, real-time monitoring, and activity logs could help track the impact of AI agents by collecting data on their activities. These data could help to inform potential interventions to prevent and respond to malfunctions.

‍

Limitations and risks of obtaining visibility

While these visibility measures have the potential to substantially improve our understanding of AI agents, they also have some limitations and pose risks of their own.

First, we have thus far only discussed visibility measures for deployers. However, users can also run AI agents themselves using cloud computing resources or their own hardware. In these cases, the burden would ultimately be on users to ensure that the agents they are running are providing appropriate identifiers. These users would also be responsible for running the real-time monitoring and activity-logging software that is needed to detect misbehaviour. However, some users may not choose to take these precautions. Self-monitoring also does not prevent cases of intentional misuse.

One alternative option in this case may be to have providers of cloud compute services implement variants of visibility measures. For example, compute providers could track especially large uses of compute and investigate whether they correspond to potentially concerning uses of agents. Additionally, providers of external tools could require proof that agents accessing their tools have implemented visibility measures. For example, a bank could require a digital signature that uniquely identifies the agent accessing its services.⁷

On the whole, though, obtaining visibility when the deployment of AI agents is “decentralised” is much more difficult. When deployers no longer have access to an agent’s inputs and outputs, it is harder to ensure that relevant information about the agent will be shared.

Second, whether or not visibility measures involve deployers, they raise significant concerns regarding privacy and abuse of power. Information about the actions of AI agents may reveal significant information about their users. Those collecting such information could abuse their power, for example to engage in excessive surveillance or the inappropriate sale of personal data. Valuable services could be barred to AI agents that did not come from approved deployers. If so, users may be locked into using agents that do not serve their best interests or meet their specific needs.

An important question is therefore how to design data collection and access practices that both provide useful information about agents and minimise the misuse of the information. One approach may be to limit visibility measures to high-stakes domains. For example, visibility measures might be a higher priority for agents that interact with the financial or legal systems. Another option may be to limit the granularity of collected information. In some cases, it may suffice to know that an agent has simply accessed a certain tool without knowing the precise inputs and outputs. 

Finally, information is insufficient for reducing risks. We also need effective processes to allow relevant actors to use and act on the information. Once AI agents and visibility measures are widely deployed, they will likely generate a lot of data. In light of the rapid pace of AI progress, delays in acting on this information could be quite costly.

‍

Conclusion

Given the potential impact of AI agents, visibility into their use is important for understanding and limiting risks. Identifying agents in their real-world interactions, monitoring their behaviour in real time, and keeping records of their behaviour could all be effective strategies for ensuring safety. 

However, more work is required to understand how, when, and to what extent these visibility measures could be best implemented. In addition to addressing potential misuse of the information generated from visibility measures, we also need efficient processes for making use of the information to inform actions. Ideally, visibility could guide the use of a wide range of strategies by individual users, deployers, and governments to manage the risks of AI agents.

‍

‍

‍The author of this piece would like to thank the following people for helpful comments on this work: Markus Anderljung, Stephen Clare, and Ben Garfinkel.

‍

Alan Chan can be contacted at

al*******@go********.ai

Blogpost 21/2024

In February, the German Federal Constitutional Court (Bundesverfassungsgericht) rejected a motion regarding electoral thresholds in EU electoral law, finally allowing for the necessary national approval of Council Decision 2018/994. This Decision intends to amend the European Electoral act and, according to Article 223 (1) TFEU, must be approved by all Member States. Up until now, the court had held that thresholds in European elections were not compatible with German constitutional law. However, a draft legislative act proposes that some Member States would be obliged to establish electoral thresholds for European elections. With this new judgement, the Bundesverfassungsgericht joins other European courts in finding thresholds to be compatible with national constitutional law.

This blog post aims to provide context for a decision that might very well change the composition of the European Parliament.

 

Previously on… electoral thresholds

In elections, citizens cast their votes in order to have their opinions represented in a parliament. In theory, representing every political view leads to a better democracy in which minority voices can gain much influence. However, fragmentation of a parliament can interfere with finding a consensus and thus hinder governability. By requiring a minimum percentage of votes a party must gain to be allocated a seat in a parliament, electoral thresholds seek to balance representation and governability. Approximately half of all Member States currently employ electoral thresholds in European parliamentary elections. The threshold is 5 percent in nine states (Czechia, France, Croatia, Latvia, Lithuania, Hungary, Poland, Romania and Slovakia), 4 percent in Austria and Sweden, 3 percent in Greece and 1,8 percent in Cyprus. Fourteen Member States do not currently have minimum requirements for allocation of European Parliament seats.

Thresholds are common in German electoral law. On the federal level, a party must gain at least five percent of votes to be allocated a seat in the German Parliament, the Bundestag (§ 4 (2) no. 2 Bundeswahlgesetz). Similarly, in the first European elections, German parties had to pass a threshold of five percent and, later, of three percent (§ 2 (6) resp. (7) Europawahlgesetz [old version]). In 2011 and 2014, the Bundesverfassungsgericht ended this practice. While it has always held that the federal threshold is not only legal, but constitutionally mandated, the Court saw clear differences between the German Parliament and the European Parliament. Governability is extremely important for the Bundestag, which is responsible for electing the Bundeskanzler (chancellor) and where the governing parties hold much power. However, on a European level, the European Parliament is not as involved in the governing and does not require a stable majority. Although the Commission President is elected by the Parliament (Article 17 (7) of the Treaty on European Union [TEU]), and the College of Commissioners can be removed by a parliamentary motion of censure (Article 17 (8) TEU), the Commission does not need continuous support from the Parliament in order to govern. For example, in the second reading during the ordinary legislative procedure, an act can pass without a parliamentary procedure when the Parliament either does not vote on a Council position or does not disapprove of the position with a majority vote (Article 294 (7) lit. a, b TFEU). Groups in the European Parliament differ from their national counterparts as well: the strongest groups do not form a ‘government’, Commissioners usually come from different political groups. Since the Parliament is so diverse in nationalities, languages, cultures, and political opinions, large groups provide a form of integration: internal debates often happen so that groups can speak with one united voice when it comes to plenary debates. Fragmentation is therefore, according to the Bundesverfassungsgericht, not as daunting on the European level as it is in the German Bundestag.

Other Member States’ Courts have also ruled on their respective electoral thresholds. The Czech Constitutional Court also argued that national parliaments and the European Parliament are different by nature and can not be held to the same standards (para. 70). However, a stable majority in the European Parliament is elemental to the functioning of the European Union (paras. 71, 72). It concluded that the European electoral threshold required by Czech law was in line with the Czech constitution. The Italian Constitutional Court also held that thresholds were compatible with the Italian Constitution as they are ‘typical manifestations of the discretion of a legislator that wishes to avoid fragmented political representation, and to promote governability’. The French Conseil Constitutionnel also ruled the electoral threshold to be in line with the French Constitution. It based its judgement on two pursued objectives: the favouring of ‘main currents of ideas and opinions expressed in France being represented in the European Parliament’ and the avoiding of fragmentation.

 

Why did the Court have to decide again?

European elections are governed by national electoral laws. A framework for these national laws is the European Electoral Act from 1976, which is drawn up by the European Parliament and adopted by the Council (Article 223 (1) of the Treaty on the Functioning of the European Union [TFEU]). In 2018, the Council voted to amend the Electoral Act and introduce electoral thresholds. According to the second paragraph of Article 3 of the Council Decision 2018/994, Member States may set thresholds of up to five percent. Constituencies comprising more than 35 seats are obliged to set a threshold of at least two percent. Only three Member States are currently allocated more than 60 seats: France, Italy and Germany. Since French and Italian electoral law already employ thresholds, this new rule would only affect Germany. In order for this Decision to come into effect though, the procedure of Article 223 (1) TFEU must be followed: Member States have to approve of the amendment ‘in accordance with their respective constitutional requirements’.

German constitutional law mandates that the national legislative bodies (Bundestag and Bundesrat) approve of the law with a two-thirds majority (Art. 23 (1) 3, Article 79 (2) of the Grundgesetz). Both decisions were reached in 2023. However, the Bundespräsident (head of state) has to sign the decision for them to come into full effect. Until this happens, the Council Decision has not been approved and the Electoral Act cannot be amended.

 

The Court’s decision

German satire party Die Partei currently holds two seats in the European Parliament, having won a share of 2.4 percent of German votes in the last European elections. Their two Members of Parliament, one of which joined the Greens/EFA group, tried to stop the Electoral Act from coming into effect by calling upon the Bundesverfassungsgericht. They argued that, as previously decided by the Court, thresholds on the European level were unconstitutional. Substantively, they stated that thresholds infringe on the right to equal opportunities for minority parties and weaken democracy (para. 29).

However, the German Constitutional Court has longstanding jurisprudence on their competence ruling on national measures in the scope of EU law and has developed three tests. The Court only tests whether an EU act is ultra vires or whether the German constitution is affected at its core (Identitätskontrolle). It does not test Union law in light of national fundamental rights as long as EU fundamental rights provide a comparable level of protection (Solange II). The petitioners argued that the Council Decision was ultra vires and that it violated the constitutional identity. The Court found that the petitioners had not substantiated this claim enough. German approval of the electoral law amendment does not confer new competences to the European level, since Article 223 TFEU already exists. Therefore, the amendment does not overstep competences and is not ultra vires (paras. 93 f.). It also did not follow the petitioners’ claim that German democracy, and therefore the German constitution, were infringed. The EU holds itself to democratic standards. Though the EU’s interpretation of democracy might differ from the German interpretation, democracy as a constitutional standard is not affected at its core when modifications are made (para. 101 f.). EU legislative bodies are awarded a prerogative to assess and shape electoral law (paras. 121 f.).

In a departure from past decisions, the Bundesverfassungsgericht now sees the danger of a deepening rift in political views, resulting in more fragmentation of the Parliament (para. 17). It now argues that a stable majority in the Parliament is essential to its important responsibilities as a legislative body equal to the Council, in the creation of a Commission and the budget power. Since the two biggest groups in the parliament no longer hold an absolute majority in the Parliament, finding this majority proves to be more challenging (para. 123). Additionally, the groups’ ability to integrate different views is limited. Preventing a more fragmented and heterogeneous Parliament is therefore a legitimate objective.

The Court therefore rejected Die Partei’s motion. As a result, the German approval of the European Electoral Act amendment can now come into force.

 

Outlook

Will electoral thresholds be applied in the upcoming 2024 elections? No. The European elections in June will still be governed by the national electoral laws that have been in effect for the past few months. Additionally, Germany was only one of two Member States still pending approval: Spain has yet to approve of the amendment. Mandatory thresholds could eventually be applied in the 2029 elections.

However, maybe future elections will be held in accordance with very different laws. For quite some time, forces inside the European Parliament have pushed for a European Electoral Regulation that would be applicable in every Member State without national legal implementation. These drafts have often included proposals for transnational lists or pan-European constituencies. So far, these proposals have always failed to win over the approval of national governments in the Council.

It seems more likely that national legislation will adapt and that we will see fewer minority parties in the European Parliament. Let us hope that stopping fragmentation in the European Parliament will be a mirror of a less divided, less extreme European society.

Blogpost 24/2024

The coming of spring promises many changes, including a newly elected European Parliament and a new college of Commissioners leading the European Commission. The re-opening of the Spitzenkandidaten system has also stirred the debate on the democratic legitimacy of the EU institutions. Focusing on the European Commission, one question that needs answering is about its members: are the European Commissioners creatures of the world of politics or instead independent experts of a technocratic ‘government’?

Looking at it from a constitutional perspective, the Commission is a unicum, with no one-to-one equivalent in nation states. The only substantive provision in the Treaties regarding the work of Commissioners is included in Article 17(3) TEU, which specifies that Commissioners shall be appointed ‘on the ground of their general competence and European commitment from persons whose independence is beyond doubt.’ However, that does not mean that Commissioners must be completely apolitical:  indeed, the Guidelines of the Commission provide for the possibility of Commissioners taking part in the campaigns and elections of the European Parliament (see Article 10).  While political standing helps to set the wheels in motion, there should also be a sense of democracy and direct responsibility to the electorate of Commissioners, if the Commission is to resemble a ‘European Government’. If priority is to be given to Commission duties over party commitment (Article 10(1) Commission Guidelines), then Commissioner candidates are hardly going to act in their neutral and professional capacity, if that would simultaneously mean kicking away the ladder that puts them in their current position. In other words, if Commissioners belong to political parties, this inherently puts them into a precarious conflict between party affiliation and their work as independent public officials (Gehring and Schneider p. 1). 

The legal framework to appoint Commissioners

Since the transformation from the High Authority and the merger in 1967, the Commission has seen a gradual increase in the number of Commissioners (from the original nine to the current 27). The Delors administration is still cited today as the ‘golden standard’ for Commission administrations. The direction and dynamism of this administration helped to solidify the position of the European Commission as the principal advocate for further integration. Among its greater achievements are the completion of the Single Market and the introduction of a single currency. The main reason for setting the Delors administration as the measuring stick is a specific attribute the administration possessed – an ability to identify the political objective, weigh up competing interests, and set out a road map to achieve it. In a sense, one could say the Delors administration was political on the EU level.

Since then, the power of the Commission has steadily increased, with Romano Prodi being dubbed ‘virtually the prime minister of the European Union’, mainly because the President of the Commission could co-decide with Heads of Government/State of the Member States on who should sit in the new administration – a change introduced with the Treaty of Amsterdam (Article 4(4)). At the time, both the German Chancellor Schröder and Mr. Prodi expressed the desire to form the new Commission as a body of independent experts and not of retired or retiring politicians. How does this reflect on the appointment of the Commission as the ‘European Government’?

Article 17(7) TEU stipulates that the candidate for President of the Commission is to be proposed by the European Council, taking into account the results of the European Elections, and then to be elected by a simple majority in the European Parliament.

For the rest of the Commissioners, neither the Treaties nor any inter-governmental agreement specifies how candidates for the Commission are to be chosen in individual Member States. In other words, no source of EU law regulates national procedures of selecting a candidate for the European Commission. The singular provision on this is Article 17(3) TEU that states that ‘the members of the Commission shall be chosen on the ground of their general competence’ and not based on their electability as politicians. This paucity of procedural guidelines itself leaves Member States free to implement their own procedures. For example, Austria regulated it partially in Article 23c of its Federal Constitutional Law, while Slovenia included it into its Cooperation in EU Affairs Act. Similarly, both examples give discretionary power to the national government to propose a candidate, who has to be approved by the national legislature – either the pertinent committee or the plenum.

The Commissioner’s role – is it political or technocratic?

The technocratic side

While it is customary for national governments to use the political apparatus to get elected, some scenarios require an appointed technocratic government of experts to lead the country, in the capacity of interim or caretaker governments (Lachmayer and Konrad). Such technocratic governments are considered to be above party politics, which enables them to bridge the political gaps between political parties.

Since the job of Commissioner requires a certain amount of independence and impartiality towards individual Member States, a technocratic candidate, with no political background, yet with expert knowledge in the department’s work, would seem to meet this ideal. If Article 17(3) TEU is to be analysed word by word, then candidates are to be ‘chosen on the ground of their general competence and European commitment from persons whose independence is beyond doubt’. While the administrations before the Juncker administration have not been viewed as ‘political’, they always included experienced public officials, who have been well acquainted with the functioning of the European Union (Peterson p. 9-21). In fact, if the principal role of the Commission is to combine all 27 different national perspectives and unite them into one voice, while reaching the optimal consensus, that ‘speaks for Europe’, technocratic – and not political – qualities seem a better choice.

While the role of Commission President has certain functions resembling a Head of Government (Craig and de Búrca, p. 32), which require a more political profile, the role of an individual Commissioner itself does not necessarily require large political capital. This makes the Commission wear ‘two hats’ (as the 19^th-century expression goes) – being involved in politics, on the one side, and remaining above the political ground, on the other. The potential problem that could emerge from a politically-disengaged administration may be the political implementation of the Commission’s work: if the Commission’s work is detached from the political reality, both sides of the spectrum – the political and the administrative – are doing Sisyphean tasks.

In the past, it would seem that almost every administration had a mixture of both. This might be attributed to the selection procedure, where Member States should (ideally) propose three candidates for the (future) President of the Commission to choose from. The last two European elections have shown us that this formal requirement is mostly ignored, even when the Member States were asked to adhere to a female-male balance of the Commission. As mentioned previously, every administration had a combination of both the administrative and the political component, but there has never been a formal requirement to balance both sides in the entire College of Commissioners. A possible reform of this is discussed below.

The political side

Some authors consider the Commission to be an inherently political institution, which sometimes tries to tone down its own political importance, to give itself a sense of impartiality. The practice of appointing party members as the candidates to become Commissioners is evidently more widespread, with 24 Commissioners being national party members or affiliated to a party. As far as political appointments are concerned, the past has also shown us that playing party politics in the Commission does not end well: as seen by the example of Sylvie Goulard in 2019 as the French candidate being replaced by Mr. Breton.

The administration under Jean-Claude Juncker was judged as one of the more politically motivated Commissions in the history of the EU. With Mr. Juncker being elected following the Spitzenkandidaten procedure, the very birth of this administration was political. When forming his Commission, he ‘promised to put together a political Commission’ (Juncker, 2014). While this might have been desired to ‘revamp’ European integration, it has proven to be a significantly damaging factor for the impartiality of the Commission on rule of law issues (noticeably in Poland and Hungary). A ‘deliberate governmental strategy of systematically undermining all checks and balances in Poland’ (Pech) and ‘saying goodbye […] to liberal democracy’ (Hungarian Prime Minister Orbán in 2018) were not developments that took place over a short period of time. The Commission certainly tired to remedy the situation (Michelot, 2019), yet showed internal splits and hesitancy in launching Article 7 TEU proceedings. Perhaps the most important setback is that a political Commission cannot ‘pretend that all of the EU’s policy goals are reconcilable and mutually supportive’ (Dawson, 2018): in the crucial politically disputed areas, a political Commission pursues the prevailing political majority and not ‘the wider EU interest’.

Taking these findings into account and applying them to the current electoral campaign, having Member of the European Parliament (MEP) candidates who already had a post in the Commission could improve a party’s credibility in European affairs as well as signal that the candidate is prepared to face public scrutiny, at least at the level of his/her local constituency. So far, at least five of the current Commissioners are also running for a seat in the European Parliament including Ursula von der Leyen and Nicolas Schmit as Spitzenkandidaten. This, of course, does not translate to immediate electoral success for their party but could be an important factor in the final vote. Standing for the European Elections could increase a candidate’s democratic legitimacy as an individually chosen representative to hold the post of Commissioner and contribute to further democratise the Commission as an institution.

Since elections are difficult to predict, national governments rarely announce their choice for the future Commissioner, nor take a stance on the Spitzenkandidaten before the results. If a governing party does announce a candidate, it is usually either someone from their own ranks or someone with close ties to them. In doing so, the party brands them with their political colours. By avoiding naming a candidate in the campaign stage of the European elections, they partly avoid the possible embarrassment if their party were to lose the election and at the same time keep their options open, in case a broader consensus would be required.

In this regard, the current campaign in Slovenia is quite intriguing. The biggest government party announced their candidate for the future Commissioner, without even having a full list of Slovenian candidates for the European Parliament. It is confirmed that their candidate Tomaž Vesel will not lead the party into the election, nor will he even stand as a candidate. Nationally, this decision has caused a governmental crisis, allowing the Government to ignore the results of the European elections already before they have even come out as well as the opinion of other coalition parties due to the opaque rules on naming a candidate for the Commission. It is difficult to comprehend how a nominee for the Commission, who neither participates in the campaign, nor even stands as a candidate for the European Parliament can help solve the democrat deficit problem in the EU.

Possible reforms – fostering more democracy in the selection procedure

As is often the case, a blend of both systems i.e. the technocratic and the political system would be the optimal solution. As the apex of the European bureaucratic machine, the Commission requires a political charge to create wider policy. However, the bigger picture requires of the Commissioners’ expert knowledge of their own department and a large amount of independence, if they intend to do a successful job. If we accept that the Commission is simultaneously a political and a technocratic institution, might it not be sensible also to try and strike a balance between Commissioners being both political actors and impartial experts, to maximise the Commission’s efficiency?

So far, no additional requirements for Commissioner candidates have been voiced, yet it would seem that several of the incumbent Commissioners have decided to actively participate in the coming European elections, standing for election as MEPs. In this light, it would perhaps be prudent to consider the long-standing British constitutional practice that ministers – the executive – are simultaneously members of the legislature. This makes the British Cabinet effectively ‘a committee of the legislative body selected to be the executive body’ (p. 48 Bagehot 1867).

This holds significant advantages in terms of democratic accountability, since all members of the executive have been directly chosen by the people to represent them in the highest democratic institution – the parliament. In other words, this enables the public to narrow the pool of possible candidates that can hold public office. It also significantly prevents the occurrence of nepotistic appointments in the executive and legislative institutions. At the same time, ministers enjoy a certain degree of independence and a high political profile, regardless of their position in government, which contributes to their independence in cases of executive autocracy. An example of this is the unprecedented revolt in the final days of Mrs. Thatcher’s government.

Many of the above-mentioned strengths would improve the current constitutional predicament of the Commission: if fostering more democracy is the goal, then requiring future Commissioners to be a part of the biggest international democratic legislative body would give the peoples of Europe far more power in choosing their own representatives as well as the country’s representative in the Commission (although the Commissioners are expressly forbidden from following instructions of national governments or other entities). Giving the electorate the power to decide who enters Parliament and consequently the Commission would also impede the search for the ‘ideal candidate’ to lead a department. Additionally, if only members of the legislature could also occupy positions on the MEP’s staff, then the unfortunate spat on President von der Leyen’s staff and the accusations of nepotism might have been completely avoided.

The incorporation of these potential changes would, however, likely only be possible by re-opening and amending the Treaty on the European Union (TEU) and the Treaty on the Functioning of the European Union (TFEU).

The epilogue after June

It should be noted that there is an important difference between participating in the European elections and being appointed as Commissioner. How one is elected (or appointed) has consequences on one’s job performance. Does participating in the elections hinder a candidate’s ability to act independently and apolitically in the future? Though the question is meant to be rhetorical, no politician would like to return to the electorate without having fulfilled at least a part of the promises and policies on which he or she was elected.

After the 9^th of June, the future administration of the Commission will start taking shape. Since the biggest political groupings have returned to the election campaign with their own candidate to lead the Commission, we can justifiably claim that the Spitzenkandidaten are back. This would effectively solidify the claim of the biggest ‘winners’ in June to demand their own candidate is nominated as the President of the Commission. Given the lukewarm reception of Mr. Juncker and the rejection of Manfred Weber in 2019, the selection of the candidate for Commission President or election of the Commission President could go either way. The selection of the President of the Commission could just as well affect the proposals of Commissioners from the Member States. It would be important however, to consider the political and the technocratic arguments and ultimately usher in more democracy to the European Commission, by creating a balance of both interests – either in terms of quality or quantity.

Blogpost 25/2024

On 18 April 2024 the European Commission issued a recommendation for a Council Decision authorising the opening of negotiations for an agreement between the EU and the UK on youth mobility. This is the first time since the signing of the Trade and Cooperation Agreement (TCA) in 2021 that the EU has proposed the conclusion of a legal framework for mobility of persons between the EU and UK. Free movement of persons ceased between the two as from 1 January 2021. Since then there has been a continuing exodus of EU nationals from the UK: 87,000 more EU nationals left the UK than came to it in 2023 (COM(2024)169 p 2). EU national students coming to the UK has dropped by 50%.

In response to this changing landscape of mobility, in 2023 the UK government has been approaching some (but not all) Member States regarding the possible negotiation of youth mobility arrangements based on existing UK national law. This unilateral action has sparked the Commission to seek a negotiating mandate from the Council to block possible bilateral arrangements between the UK and some Member States to the exclusion of others. This is consistent with the Council position adopted on 23 March 2018 that any future partnership between the EU and the UK on mobility of persons should be based on full reciprocity and non-discrimination among Member States.

As a result of the upheaval which the decision to leave the EU caused to the UK political class, including among other things a change of prime minister, while the UK had been interested in youth mobility in 2018, by 2019 the government was no longer willing to include this in the TCA. This has meant that youth mobility between the two has been regulated by national law in the UK and by a mix of EU and national law in the Member States. The UK has a long standing youth mobility programme limited to young people, nationals of countries specified in the immigration rules, between the ages of 18 to 30 or 18 to 35, depending on what country the person is a national of, and limited to two years. No EU country is included in this category (though Andorra, Iceland, Monaco and San Marino are).

The Commission proposes that a new youth mobility agreement be part of the TCA framework and remains neutral on whether it would be a Union-only or mixed agreement, something to be determined at the end of the negotiations. Similarly, it considers that the legal basis for the agreement would have to be determined only at the end of the negotiations. Neither of these issues is likely to meet with enthusiasm by the Council which may wish a clearer remit to the Commission regarding what can be negotiated. The Commission considers that only a formal agreement between the UK and the EU will achieve the objective in providing legal certainty and addressing the issue of non-discrimination. It states that only a “binding mutual understanding in the form of a formal international agreement” can guarantee legal certainty. Nonetheless, the Commission envisages that the agreement would be supplemental to the TCA and would be part of its single and uniform institutional framework, including rules on dispute settlement.

For young people in the EU and the UK this would be a rather unsatisfactory framework on account of Article 5 TCA. This states that (with a sole exception for social security) “nothing in this Agreement or any supplementing agreement shall be construed as conferring rights or imposing obligations on persons other than those created between the Parties under public international law, nor as permitting this Agreement or any supplementing agreement to be directly invoked in the domestic legal systems of the Parties.” So young people seeking to exercise mobility rights under any new agreement would not be able to rely on such an agreement if it is adopted within this framework. This could only be resolved if Article 5 were also amended to exclude from its scope not only social security but also youth mobility.

The Commission proposes that the scope of the agreement would cover twelve issues. First, the personal scope would be limited to EU and UK citizens between 18 and 30 years. The period of stay would be four years maximum. There would be no purpose limitation on mobility, young people could study, work or just visit if they want to. There would be no quota on this category. The conditions applicable to the category should apply throughout the individual’s stay. Rejection grounds would be specified. The category would be subject to a prior authorisation procedure (ie specific visa to be obtained before arrival). For UK citizens, their mobility would be limited to the one Member State where they had received authorisation (leaving open the question whether the periods for be cumulative or consecutive in different Member States). Equal treatment in wages and working conditions as well as health and safety rules must be respected on the basis of non-discrimination with own nationals. This may also include some aspects of education and training, tax benefits etc. In particular, equal treatment as regards tuition fees for higher education is planned. This would mean that EU students seeking to study in UK universities under the youth mobility scheme would only pay home student fees which are dramatically cheaper than overseas student fees which are currently applicable. Interestingly, the Commission proposed that this home student fee provision should apply to all EU students in the UK including those who arrive on student visas rather than youth mobility ones. The UK’s ‘healthcare surcharge’ would also be waived for this category. Finally, the conditions for the exercise of family reunification would need to be specified.

The Commission plans that any youth mobility scheme should be without prejudice to other legal pathways for migration and EU rules on permanent or long-term resident status.

For the EU, such a youth mobility scheme between the UK and the EU would add to an already rather complex field of EU competences. The Students and Researchers’ Directive covers conditions of entry and stay for the purposes of research, studies, training, voluntary service, pupil exchange schemes or educational projects and au pairing. This would certainly cover quite a lot of what is planned for youth mobility. However, the Commission appear not to be keen on using Article 79 (2) (a) and (b) TFEU, the basis of that directive for the purposes of this initiative. One of the reasons is that all the categories of persons covered in that directive need a sponsor (which could be a university, an employer or a training institution) within a Member State who is saddled with a variety of obligations regarding the third country national to ensure that they comply with general immigration conditions. Such a sponsorship approach is not intended by the Commission for UK-EU youth mobility. Further the Commission’s objective is to achieve reciprocity between the parties and non-discrimination among the Member States and their nationals. This is not an element of the directive. Thus, a new agreement seems to be the preferred approach – the Commission appears to prefer the ‘free movement’ approach rather than the sponsored one. Yet, as mentioned above, if the objective is to provide legal certainty to Europe’s young people regarding moving between the EU and the UK, the TCA does not seem to be an appropriate tool either as it specifically rejects that legal certainty by denying the right to individuals to rely on its provisions before the authorities or courts of the parties.

At the time of writing, it is unclear how the Council will approach this proposal. There are indications that some Member States may not be enthusiastic (Hungary is one) worrying that their skilled young people may be enticed to go to the UK rather than staying at home. But the majority appears to be very positive towards any move to normalise mobility between the two parties.