Artificial intelligence and European identity: the European Commission’s struggle for reconciliation

ABSTRACT The European Commission (EC) has recently developed its AI strategy which aims to develop and mobilize for a European version of AI. This article seeks to generate an in-depth understanding of how the EC frames its AI version as distinctively European. A frame and content analysis methodology is used for understanding how the EC struggles to reconcile AI and European identity. European identity consists of European historical narratives, different societal and political purposes and values. It is found that the EC seeks to legitimize the European AI alternative through beneficial outputs and narratives based on historic technological leadership and that via framing practices the EC tries to foster European integration. This paper concludes that the EC's communication efforts are inconsistent and might not be sufficient to generate trust in AI among the European public.


Introduction
Language in politics matters.Via a particular framing that resonates with cognitive paradigms, such as identities, politicians can direct society's attention, can mobilize citizens and stakeholders and influence how we perceive a topic despite facts and data (Druckman, 2014;Wehling, 2020).This is also true for the so-called 'AI revolution' that allegedly will reshape entire societies (Bareis & Katzenbach, 2022;Köstler & Ossewaarde, 2022;Ossewaarde & Gülenç, 2020).Yet, there is a lot of uncertainty about whether AI is a blessing or a curse (EPSC, 2018;Gibbs, 2014).Various scholars have argued that the embeddedness and fit of technologies like AI into societal value and belief systems is vital for its acceptance; and that framing AI as a positive force for realizing something good is a useful technique for generating this resonance (Druckman & Bolsen, 2011;Geels & Verhees, 2011;Lempiälä et al., 2019;Radu, 2021).The European Commission (EC) has recognized this.It emphasizes that '[to] address the opportunities and challenges of AI, the EU must act as one and define its own way, based on European values, to promote the development and deployment of AI' (European Commission, 2020f).The EU aims to maintain 'technological leadership' (European Commission, 2021e, p. 1) to secure European values and to '[…] help shape global norms and standards and promote trustworthy AI that is consistent with Union values and interests' (European Commission, 2021e, p. 5).By referring to European values the EC presents its AI strategy as part of a larger debate on what constitutes a European identity (EI)what defines Europeanness or a European way of life.This becomes clear when the EC distinguishes between its European version of AI and other political variants of AI, such as those of the US, Russia or China, that allegedly contain threats to European values (Roberts et al., 2021).If AI is framed in a matter that is in line with values and narratives from a collectively shared EI, salience among the public can be enhanced, the argument goes (Hänggli, 2012).This sensitises our gaze for critically observing how the political-administrative shaping of the European version of AI is accompanied by the EC's political communication; which ideational frames are presented by impactful EU policy-makers.The EC is a driver of European integration.It has the exclusive right for legislative initiative and avid activity in stakeholder engagement processes.It is a key factor in shaping the European AI Strategy and European societal evolution and is therefore the object of investigation in this article.
This article aims to illustrate how the EC seeks to reconcile its AI program and its longstanding EI politics in light of the ambiguity of technological revolutions.To do so, this article entails an analysis of the EC's EI frames.It focuses specifically on recognized threats of AI to European values (such as non-discrimination, democracy and freedom) and to Europe's alleged competitive position in the booming global AI technologies market.While the framing of AI has been analysed in national policy frameworks and newspaper discourse (Bareis & Katzenbach, 2022;Garvey & Maskal, 2020;Köstler & Ossewaarde, 2022;Liebig et al., 2022;Ossewaarde & Gülenç, 2020;Radu, 2021;Sun et al., 2020), there are so far no studies analysing how the European institutions try to do so.Scholars have recently addressed the content and development of EU AI policy documents (Justo-Hanani, 2022;Vesnic-Alujevic et al., 2020), recognizing the impact of European involvement on AI development in the context of global competition (Monsees & Lambach, 2022;Tinnirello, 2022;Ulnicane, 2022).Ulnicane (2022) has analysed EU institution's AI programs from a normative power perspective.Detailed scrutiny of how AI programs and political communication around these programs try to reconcile EI and AI is however missing.The insights generated by filling this gap are both of scientific and societal relevance.The communication of the European AI program has to address the concerns of stakeholders and citizens regarding AI.The EC hopes to generate public trust by presenting an AI program in line with European values (European Commission, 2020g).Gaining more insights into what a connection of EI and AI entails, will give societal actors a deeper understanding of what they can expect from a European version of AI.This will also constitute the framework, the societal contract, to which the public will hold AI policy makers and companies accountable.Examining how the EC frames a European version of AI will contribute to the discussion on potential pitfalls and successes of the AI program.This knowledge is relevant for conducting research on European integration in a digitalized society that is shaped by an identity politics of technology.
This article proceeds as follows.We first provide the EC's definition of AI and describe the global context in which the EU's AI strategy evolves (2.1).We then provide a definition of EI and EI politics (2.2) and the concept of framing (2.3), which constitute the theoretical foundation of our analysis.In our methods section we translate this theoretical debate into a coding scheme.Our content and frame analysis reveals how the EC seeks to reconcile its AI strategy and its EI politics.We present three dominant clusters of frames (4. 1, 4.2, 4.3) and exemplify the construction of the identity frames in a table.In a last step we analyse whether the EC uses othering to differentiate its European version of AI from other international actors (4.4).Our analysis illustrates how the EC presents AI as a threat to a collective identity and, yet, also as a necessity for delivering on the promises of the EI.We find that the EC tries to reconcile AI and European values through framing AI as trustworthy and humancentric.The EC emphasizes its wish to safeguard European values and identifies its European version of AI as catalyst of European integration, well-being and prosperity; the EC downplays the risks of how its push for AI may contradict its EI politics.

The context of global AI trends
One of the challenges in AI discourse is the lack of a common definition of AI.To allow a consistent analysis of the EC documents, this paper follows the definition introduced by the EC in their Communication 'Artificial Intelligence for Europe'.Accordingly: Artificial Intelligence (AI) refers to systems that display intelligent behaviour by analysing their environment and taking actionswith some degree of autonomyto achieve specific goals.AI-based systems can be purely software-based, acting in the virtual world (e.g.voice assistants, image analysis software, search engines, speech and face recognition systems) or AI can be embedded in hardware devices (e.g.advanced robots, autonomous cars, drones or Internet of Things applications) (European Commission, 2018d, p. 1).
The European Union (EU), like the United Nations (UN) (AI for Good, n.d.) and the Organisation for Economic Cooperation and Development (OECD) (OECD, n.d.), emphasizes the importance of ensuring trustworthy AI systems.The EC stresses that […] an EU strategic framework based on fundamental values will give citizens the confidence to accept AI-based solutions while encouraging businesses to develop them (European Commission, n.d.).
Other global leaders in AI technology, such as China, the United States or Russia, tend to follow different approaches towards AIapproaches that have the potential to clash with the European vision of safety and trustworthiness.In the USA, the government plays a comparably weak role in shaping AI policy and investment in AI.Instead, leading Silicon Valley firms dominate the American development of AI and have a track record of showing little commitment to values like privacy, democracy or the rule of law.The US prioritises national competitiveness in AI development not the safeguard of individual rights (Craglia et al., 2018;Roberts et al., 2021).In the Chinese AI program, AI-related technologies are meant to maintain social harmony and control; for instance, through the further sophistication of the so-called 'social credit system' which provides a rating of citizens' reputations (Craglia et al., 2018).According to the EC's Joint Research Centre (JRC), the absence of public criticism of AI systems undermines the process of developing trustworthy AI systems that ensure the inclusion of moral responsibilities, transparency and accountability, as perceived by the EU (Craglia et al., 2018).In the full awareness that AI systems can violate key European values like freedom, democracy, rule of law, justice and human rights, the EU has recognized the need for developing a version of AI that reflects a EI; and is the basis for a global image as a standard-setter and moral leader in a global politics of AI (Calderaro & Blumfelde, 2022;Craglia et al., 2018;EPSC, 2018;Monsees & Lambach, 2022).But what does this European self-image look like?What constitutes an EI to which other approaches towards AI are dissonant and to which a European version of AI must resonate?

The three branches of European identity
European identity politics has been a constant in European integration discourses since the establishment of the European Economic Community (EEC) (Bee, 2008;Börzel & Risse, 2018;Caporaso & Kim, 2009;Kuhn, 2019;Stråth, 2002).For long, scholars have recognized the importance and existence of a EI for cultural, social and political purposes of European integration, yet, its precise meaning, definition, content and political-administrative implication is contested (Börzel & Risse, 2018;Kaina & Karolewski, 2013;Kuhn, 2019;Stråth, 2002).EI is typically considered as something exclusive, derived from a collective way of life that ultimately originates in political values (freedom, democracy, political justice, etc.).According to Stråth (2002), in European history, 'othering' has been a practice to demarcate Europe from its others.It has been a central practice in the making of shared identityfrom Charlemagne's Christian Europe to the EU's identity politics (Ossewaarde, 2013;Scheuer & Schmitt, 2009).Today, the EU enacts a EI to be able to meaningfully distinguish itself from other actors, such as the USA, Russia, China, or the UK after Brexit (Català Oltra, 2021;Lichtenstein & Eilders, 2019;Stråth, 2002).
To identify the characteristic elements of an EI Bruter (2003) distinguished two dimensionsa civic identity and a cultural identity.The former determines the degree to which one feels as part of the political system with the European institutions as actors whose policy actions influence one's daily life while the latter refers to the identification with a human community.These entities are shaped by notions of (1) a common history or traditions and (2) political/constitutive or moral values (Bee, 2008;Bruter, 2003;Català Oltra, 2021).Values represent the symbolic set of rules and normative content to which members of the group (such as the European community) commit and create expectations and recognition towards members of the group (Abdelal et al., 2006).Political documents such as the Declaration on European Identity (1973), The Union of Values in the Maastricht Treaty (1992), the Copenhagen criteria (1993) and the Constitutional Treaty draft (2003) typically consider democracy, human rights and the rule of law as fundamental political values of the EU (Abdelal et al., 2006;Akaliyski et al., 2022;Bee, 2008;Ossewaarde, 2013).In the recent EC's Promoting our European Way of Life program (EWoLp) (2019), the EI is built around the values of solidarity, equality, tolerance, non-discrimination, fairness and the European social model (European Commission, 2021b).When a social or political community is defined (3) a purpose needs to be attached to it (Abdelal et al., 2006) or specifically, a description of what kind of community it is.They represent a simplified idea of EU history, current ambitions and future goals (Lichtenstein & Eilders, 2019).Examples of this have been the purpose of peace keeping or the economic advantages of the single market (Lichtenstein & Eilders, 2019), building resilience or ensuring consumer protection (European Commission, 2020e).These three branchesreferences to European history or traditions, European values and EU purposes are the core of our concept of EI from which we extract the Commission's understanding of EI in the context of its own politics of AI.Via framing, the EC can enact such notions of an EI in its AI program.

European identity frames to enhance the resonance of AI
Framing is based on the idea that practices of sense-making of the world are not merely based on facts and data analysis, but, more importantly, are based on predispositions, such as identity, values and beliefs (Druckman & Bolsen, 2011;Wehling, 2020).To frame is 'to actively construct the meaning of the reality in question' (Hänggli & Kriesi, 2010, p. 142), by leading public attention to certain aspects, perspectives and interpretations of an issue (Hänggli & Kriesi, 2010).According to Entman (1993), framing can be defined as '[to]select some aspects of a perceived reality and make them more salient in a communicating text […]' (Entman, 1993, p. 52).Salience can be achieved if certain aspects are continuously repeated, and especially, if the content of a frame resonates with existing cognitive structures (Benford & Snow, 2000).Resonance occurs if a frame is congruent with the everyday experience of the target audience, or if the frame resonates with the audiences' cultural narrations such as myths (Benford & Snow, 2000).The EC's EI frames typically function to reflect such aspects of resonance, as these are based on cultural narrations from European history or traditions and shared values and beliefs or purposes from a civic or cultural identity.EI frames based on references to group belonging and self-understanding, and with attributions to certain values and historic moments, are important for the legitimization of political-administrative authority and European policy programs (Akaliyski et al., 2022;Bruter, 2003;Kuhn, 2019;Scheuer & Schmitt, 2009).
This means that EI frames can be a powerful tool in the hands of the EC for presenting and promoting its AI program, and for legitimizing its political-administrative role as a supranational institution.We expect that the EC uses EI frames to present its European version of AI.We expect that the EC attributes elements of EI (history or traditions, values, and political or ideational purposes) to AI trying to reconcile AI and EI; and that by framing such elements as distinctively European it enacts 'othering' in the construction of non-European or even anti-European AI.And we expect that EI frames function to promote the EC's AI program and to legitimize its supranational political-administrative authority.

Methods
To find out how the EC seeks to reconcile its AI strategy and its European identity politics in its framing practices, we collected three types of EC documents that were released in the period of 24 April 2018 until 3 May 2021.These include nine strategy documents (Communications) from the 2019-2024 policy priorities of 'Excellence and trust in artificial intelligence'; 71 speeches and 67 press releases.A communication is usually issued by the EC when it faces a new policy challenge.It typically functions as an aspirational document, similar to a memorandum, a proposal for a policy approach or a strategy paper (Figueira, 2017).Press releases and speeches concerning the European version of AI best represent the EC's political communication efforts.Frames have proven to be more influential and credible if they have been introduced by prominent institutional speakers (Benford & Snow, 2000;Hänggli & Kriesi, 2010).Thus we selected all speeches by the leading figures of the EC, which are the President, Vice Presidents and the Commissioners, containing the keyword 'artificial intelligence'.Documents by the Joint Research Centre (JRC) were excluded because we are exclusively interested in the EC's enactment of its European alternative to AI.A complete list of analysed documents is provided in the Appendix.
Our content and frame analysis follows three steps.First, to develop an understanding of the EC's definition of the EI, and to identify EI elements in the gathered documents, we conduct a content analysis that is based on the principles developed by Phillip Mayring. Mayring (2015) emphasises the importance of setting rules that guide the analysis in advance defining individual steps beforehand while simultaneously having room to add new rules continuously throughout the analysis process.This way, the description and interpretation of the political communication content is not arbitrary but is based on tested rules and a structured and comprehensible procedure.We followed this strategy by developing a coding scheme (Table 1) that includes categories and cues or keywords that are deduced from our theoretical discussion of the EI.Our coding units are not limited to keywords, but include sentences or paragraphs with EI reference, because purposes and history can be elaborated not only in one keyword but arrive from the particular meaning of a paragraph.The specific content of the categories are extracted from the data through in-depth reading and examination of the gathered documentation, searching for elements of the categories as the content of collective identities is extremely variable while our analysis focused on the EC's AI program.
Second, we conduct a frame analysis, by analysing the sentences and paragraphs that include references to the EI and AI, to understand how the EC reconciles its AI program and its EI politics.EI frames are indicated by statements on general objectives, values and/or historic/traditional aspects of the EU in the context of AI (Lichtenstein & Eilders, 2019).We analyse how different EI keywords and cues centre around different lines of statement and argument and are, thus, clustered to form a specific frame.These EI-AI-constructs reflect the frames used to communicate a European version of AI.
Third, to develop an understanding of how the EC frames its version of AI as distinctively European, we focus on the expected EC's practice of 'othering', as a means of distinguishing a European version of AI from its non-European (mainly American, Chinese and Russian) counterparts.This way, we seek to uncover the EC's claims towards the distinctive Europeanness of the AI program, as its practices of 'othering' make elements of an EI explicit.

Analysis
In light of the 'global race for AI', the EU, in its struggle to become a global AI leader, faces the challenge of presenting an approach to AI that is legitimate, that is, an approach that is in line with its longstanding EI narrative.To do so, the AI program needs to be in line with EU purposes, history and values from the EI.In the context of its AI program, values, purposes and historical narratives based on EI are frequently addressed to reconcile the European version of AI with the EI understanding of the EC.This section presents the main frames that were extracted from the analysis, providing further insides on the EI narrative of the EC.

Using European AI for good
[T]he use of AI can affect the values on which the EU is founded and lead to breaches of fundamental rights, including the rights to freedom of expression, freedom of assembly, human dignity, non-discrimination […] protection of personal data and private life, or the right to an effective judicial remedy and a fair trial, as well as consumer protection.(European Commission, 2020f, p. 11) This quote from the EC's White Paper on AI illustrates that the EC acknowledges that if not handled responsibly and carefully, AI can pose threats to fundamental EU rights and values.The EC emphasises that a European version of AI shall prevent this from happening.Accordingly, when presenting European AI, the EC adopts a positive framing.It is a framing of using AI for good, for making people's lives better in Europe and around the globe.This universal commitment to AI, and promise of AI benefits, is illustrated by the EC's claim that AI contains the power to solve all sorts of problems, such as containing health care costs or fulfilling the UN Sustainable Development Goals (European Commission, 2020f).This is aligned with narratives presented by other global governance actors such as the OECD or the UN (AI for Good, n.d.; OECD, n.d.).The EC draws an immediate connection between AI and the fulfilment of its policy purposesfulfilling those purposes allegedly safeguards its political-administrative legitimacy.
[An] accelerated development and deployment of advanced and trustworthy AI in Europe is a pre-condition for Europe's future competitiveness and prosperity.(European Commission, 2021a, p. 9) Similarly, the EC claims that Europe needs AI to solve its problems, to strengthen its economy, to maintain Western European living standards, to enhance European wellbeing and that of future generations (European Commission, 2020d, 2020f, 2020j, 2021a).This is a powerful frame because it is connected to the lived reality of Europeans, resonating with perceptions of hopes and everyday challenges.We therefore call this dominant, multifaceted frame the 'Using AI for Good' frame or specifically, 'AI to achieve EU Purposes' (see Table 2).A good illustration of this frame are also speeches in which AI solutions for better cancer diagnoses are used to promote the AI program: 'as we speak [AI] is saving lives […]' (European Commission, 2021f, p. 1) (European Commission, 2020a, 2020k).Von der Leyen continues here by presenting AI as key for tackling climate change (European Commission, 2020c, 2020i, 2020k) which, according to a 2021 Eurobarometer survey, citizens regard as 'the single most serious problem facing the world' (European Commission, 2021g).Thus, via the 'Using AI for Good' frame, the EC informs its audience that AI is an opportunity to prevent suffering and loss of life and the environment and directs attention to this line of argumentation.In other words, not adopting AI would generate unnecessary suffering and uncertainty, which would be cruel.
With its 'Using AI for Good' frame, the EC claims that AI contributes to the reduction of human suffering.However, in the way they cluster this frame, they don't mention how technological revolutions as such come with suffering.The early industrial revolutions Leading AI means leading the so-called 'AI Revolution' and shaping the alleged 'Age of AI', just as the British Empire once led the earlier industrial revolutions and shaped industrial capitalism (Ossewaarde & Gülenç, 2020).We call this facet the 'AI Leadership' frame (see Table 2).The EC seeks European technological leadership to make the alleged 'Age of AI' a European (rather than a Russian, Chinese or American) age; and, thus, by presenting a European approach to AI that is in line with European values, the EC stresses that Europe must lead the so-called 'AI revolution' (European Commission, 2018c).AI leadership is presented as inherent part of the EI itself.
Frames based on these historic narratives in support of 'Using AI for Good' mask the risks and threats to the European valuesthe potential AI for Evil.They reconcile AI and the EI but disregard the ambiguity of past technological revolutionsand the human sufferings and destructions of nature that came with such revolutions.Such frames promote European AI, emphasising that AI will lead to something good.However, recent research on the EU's AI strategy in this context has found that 'the extent to which the EU's vision of a Good AI Society is being fully achieved is still questionable (Roberts et al., 2021, p. 68)'.

Trustworthy and human-centric AI
To overcome the ambiguity of AI, which emerges from its character as both a threat to and a necessity for sustaining the EI, particularly using AI for good, the EC tries to communicate a version of AI that is in line with European values.This is a challenge.While it comes more naturally to reconcile AI with EU history and EU purposes (see above), AI as a technology has the potential to threaten EU values (European Commission, 2018e, 2019c, 2020f) (see Table 2).One of these values is the guarantee of fundamental rights which, according to the EC, shall be core of the European version of AI (European Commission, 2020f).In their 2021 proposal for AI regulation, the EC provides the list of fundamental rights considered in its AI program (European Commission, 2021e, p. 24): Furthermore, the EC adds children's rights and the right to environmental protection.Other values that are named frequently are democracy, non-discrimination, diversity and transparency (European Commission, 2019b, 2020k, 2021a, 2021f).The EC emphasises that with the lack of European leadership on AI, AI is likely to endanger European values, specifically these fundamental rights ('AI Leadership' frame): By striving towards human-centric AI based on trust, we safeguard the respect for our core societal values and carve out a distinctive trademark for Europe and its industry as a leader in cutting-edge AI that can be trusted throughout the world.(European Commission, 2019b, p. 9) This quote illustrates that for the EC, to sustain European values AI has to be 'trustworthy' and 'human-centric'.Both these keywords are mentioned extremely frequently in the EC's AI program and are presented as main features of the European version of AI.They are also guiding principles for AI initiatives around the world (OECD, 2019).Both gain frame character through extensive repetition, rather than through connection to European values and narratives.The framework documents address the practicalities of how they try to ensure that European AI will be trustworthy.This includes the Ethics Guidelines for Trustworthy AI to safeguard fundamental rights (European Commission, 2019b, 2021a) and their risk-based approach that seeks to regulate 'high risk AI systems' suggested in the White Paper on AI presented in their 2021 proposal for AI regulation.(European Commission, 2020f, 2021e).
The EC claims that trustworthiness assures a commitment to the European values; for instance, through a liability framework, non-discrimination measures, or maintaining human autonomy (European Commission, 2019a, 2019b, 2020f, 2021f).However, in the EC's communication of the AI program, 'trust' is more often left undefined and functions as a reassuring and repetitive claim that European AI will be trustworthy and that trustworthiness entails a commitment to European values.Trustworthy functions as standalone keyword that allegedly marks 'good AI'.
When it comes to its keyword of 'human-centric AI', the EC suggests that 'people' must be at the centre of European AI development: According to the EC, European AI first and foremost must be understood as a tool to serve the people and increase human well-being (European Commission, 2018d, 2019d, 2021c).The EC does not mention how 'human-centric' AI comes to amount to a European AI.The EC discusses 'human-centric AI' in the context of European values, but without a clear relationship to the EI.Instead, 'human-centric AI' is presented as a tool for realizing EU policy objectives.An illustrative example of this is the purpose of enhancing people's well-being.The EC states that: The European AI strategy and the coordinated plan make clear that trust is a prerequisite to ensure a human-centric approach to AI: AI is not an end in itself, but a tool that has to serve people with the ultimate aim of increasing human well-being.(European Commission, 2019b, p. 1) The EC claims that human well-being is an effect of 'human-centred AI'.It communicates that human well-being is an indicator that the European version of AI is in line with European values.Such a claim is, of course, highly contestable, given that high levels of well-being, like economic prosperity and good health care systems, may well be achieved through despotism, without commitment to democracy, freedom, rule of law, justice and human rights (Helliwell et al., 2019).Furthermore, while the Ethics Guidelines for Trustworthy AI are often presented as backbone of the EU's approach to humancentric and trustworthy AI, ethicists part of the High-Level expert group developing and advising on these guidelines have voiced scepticism towards the implementation of ethics in the Guidelines itself but also the White Paper on AI: 'Ethics has lost all importance for the Commission, it seems' (Coeckelbergh & Metzinger, 2020).
Through constant repetition of its two keywords, the EC claims that 'trustworthy' and 'human-centric' AI are distinctively European but it fails to make an argument for this claim, and it fails to connect 'trustworthy' and 'human-centric' AI to specific European values (see empty categories in Table 3).In other words, 'trustworthy' and 'humancentric' are used as shiny keywords or even hollow 'plastic words' (Bourne, 2019, p. 115) which for the EC are positively connotated and suggest an imperative of cultural resonance but actually remain superficial, not connected to something particularly European.Values such as non-discrimination, transparency, or equality are important in the EC's communication efforts, but, they seem to be haphazardly employed to support an imperative of trustworthiness and human-centrism.When communicating on AI, the EC uses an EI frame to present its version of AI by attributing elements of EI to AI, but not in a way that we expected.Rather than the European values, it is EU policy purposes, like generating well-being and prosperity, that play an important role in the EC's presentation of the European version of AI.The frame of trustworthiness and human-centric AI functions as channel to promise the compliance with European values.Even though at this stage of development, it might be far-fetched to guarantee trustworthiness and ethical compliance.The imperative to act and the legitimacy for developing a European version of AI despite these threats towards EU values are generated predominantly through frames highlighting that AI development is aligned with the purpose and history branch of EI.

AI demands European integration
In the context of its AI program, the EC uses EI frames for unifying European citizens and Member States; and thereby for promoting European integration beyond policy integration.This is in line with past research on emerging technologies: 'Concerns about the EU lagging behind the United States (US) and [Asia] in science, technology and innovation have been a major driving force for European integration in research and technology policy since its early stages in the 1950s' (Ulnicane, 2022, p. 255).By emphasising the difficulty and severity of implementing a European version of AI which safeguards a European way of life, the EC calls for coordinated action (European Commission, 2018b, 2020f).We call this the 'AI demands Unity' frame (see Table 4).The EC suggests that AI requires from EU Member States that they work together to unite, rather than following their national AI trajectories of national identity politics (which is what national AI strategies tend to do (Ossewaarde & Gülenç, 2020)):

AI demands Unity
Investment in AI needs to come from the public and private sector.
Stakeholders from all parts of society need to be consulted for the EU AI framework.
We need a coordinated approach on AIto counterbalance fragmentation and to ensure a functioning single market.
To shape the global debate on AI; to be globally competitive; to become a leader in trustworthy AI, the EU has to act as one.
Unclear safety and liability provisions generate uncertainty for the market.The introduction of national initiatives risks to endanger legal certainty, to weaken citizens' trust and to prevent the emergence of a dynamic European industry.(European Commission, 2020f, p. 2) The EC highlights how AI may contribute to the further development of the EU's single market but it emphasises that successful AI deployment is only possible when Member States unite at the European level.An otherwise fragmented single market could create obstacles for companies (European Commission, 2020f).This claim addresses specifically corporate stakeholders that have a strong interest in a consistent legal framework on AI, but who also have the resources to boost the investment levels in AI.AI's uptake depends on them as, according to the EC, public-private partnerships are needed to generate sufficient investment levels that go beyond what public investments alone can administer (European Commission, 2020f).The same applies to European academia and research institutes who are asked to pool their resources and knowledge to develop excellent AI technology in unity (European Commission, 2018e, 2020h, 2021a).This way, European AI is presented as an example of cross-national aspiration and success representing the advantages of joint forces (European Commission, 2021d): '[This] goes beyond what any single Member State can achieve' (European Commission, 2020f, p. 5).By addressing different societal actors, this frame does not only function to promote policy integration but also societal integration.In this context, the EC states that it wants to make the process of developing a European version of AI inclusive and democratic, through seeking dialogue, feedback and criticism of different stakeholders and citizens (European Commission, 2018b, 2019a, 2019d).And it stresses that such European governance is needed to effectively embed AI in European society: '[L]ike any other technology, AI is not just imposed on society' (European Commission, 2018d, p. 12).Governance mechanisms like dialogue are mentioned in the EC's AI program to ensure that both companies and citizens benefit from AI; and will be equipped to use AI technologies and able to understand and reflect on the new technological revolution accruing through AI (European Commission, 2018b, 2018d).
Via the 'AI demands Unity' frame, AI is communicated as a European governance phenomenon that needs the input of diverse stakeholders.AI and EI are reconciled as the EC suggests that the process of developing European AI is based on core values such as solidarity, democracy and inclusionlegitimizing the policy process.This framing, which is dominant in the EC's AI program, suggests the importance of the EU governance level to coordinate the development and regulation of a European version of AI; thereby, the political-administrative authority of the EC is legitimized.This frame not only serves to enable the EC to mobilize diverse European stakeholders and citizens for an AI uptake to seize the alleged benefits of AI, and to safeguard the EI in the face of fierce global competition but, that the AI revolution offers a chance for the EC to foster European integration through its European politics of AI.To seize the advantages of AI and to remain sovereign, the EC claims that EU Member States, citizens and European stakeholders need to unite, work together and engage in a dialogue, expertise and education.Thereby, the integration of Member States, market integration, but also solidarity among European citizens is allegedly fostered; and the legitimacy of the European governance level secured.

A distinct European AI?
By enacting EI frames, the EC seeks to promote its European version of AI, mobilizing European stakeholders for AI's uptake.This mobilization does not occur in a vacuum out of the basic motif to seize the benefits of AI, but, instead, is strongly connected to the EU's endeavour of being a competitive party in the global race for AI leadership ('AI leadership' frame).The EC stresses the desirability of aiming for AI leadership, because, without such leadership, Europe would lose the economic and technological power allegedly necessary for upholding European values.Without AI leadership the EU will risk losing (technological) sovereignty and autonomy by becoming dependent on the applications by competitors (European Commission, 2018a, 2020b, 2020k).In the context of this 'AI leadership' frame, the EC emphasizes the unique character of its European version of AI in contrast to its rivals for AI leadership.As the EC puts it: 'What sets Europe apart from competitors is the fact that our values come first.Human beings come first' (European Commission, 2019c, p. 1).
In the EC's AI program, China and the US in particular are mentioned as competitors that are currently at the forefront of AI investment and technology (with Europe lacking behind).The EC addresses the transatlantic cooperation between the USA and the EU in the making of the AI revolution.In a context in which Silicon Valley tech oligarchs frequently violate the European values, the EC calls for the US to 'Europeanise' the AI revolution, in the sense of making AI 'people-friendly' (European Commission, 2019c, p. 2) in respect to the rights of and well-being of citizens (European Commission, 2019c).The EC emphasizes that these values are Western after all.Although hostile Silicon valley oligarchies currently dominate the American AI version, there is no distinction between the American and European alternative to AI (European Commission, 2019c).For the EC, this is different regarding the Chinese version of AI, which seems to rule out cooperation as it has no concern for European values.
Instead of specific othering, the EC emphasizes that it is open for cooperation with whoever wants to commit to European values.As the EC stresses when communicating to its partners: 'European technological sovereignty is not defined against anyone else, but by focusing on the needs of Europeans and of the European social model' (European Commission, 2020e, p. 3).This two-sided means of communication suggests that a European version of AIwith respect to the European social model and hence organized capitalismis more an inward-oriented endeavour serving the main purpose of enhancing European citizens' wellbeing and social solidarity.If the 'Using AI for Good' framing is taken up by the EC's audience, it has the potential of legitimizing the EC's political-administrative authority while mobilizing for AI's uptake alike.Yet, the frame also shows a certain weakness, given that the EC's communication efforts appear highly inconsistent: On the one hand, the EC claims singularity with its human-centric and trustworthy approach based on European values and presents its approach to AI as a solution to the global threatening of its AI frameworks.On the other hand, it emphasises and seeks international allyships based on its European values, such as with Japan, Canada or Australia.It does not specify why its approach is uniquely European.

Concluding remarks
The European Commission tries to generate trust for AI via strategic framing.Its frames are predominantly directed towards a sceptical European public that needs to be united and mobilized for AI's uptake, for shaping a European 'AI society'.In its political communication, the EC recognizes that the global politics of AI and AI technologies as such have the potential to clash with what it understands as European.Its answer to these observations is the persistent and eager repetition that the European AI alternative will be 'trustworthy' and 'human-centric'.By claiming that European AI will be based on European values, that its European version of AI is a continuum of a distinctively European history and will serve European purposes, resonance with a collective EI shall be generated.Europeans are reminded of the importance of the EU as the guardian of their collective identity and European way of living.and their shared values, historical success and today's benefits.While the EC frames European AI as trustworthy and human-centric and does so via extensive repetition of these two keywords, it fails to connect its claims to specific European values or the European social model.Important values such as democracy are not often mentioned in major recent documents such as the EC's proposal for an AI regulation (European Commission, 2021e).
Via framing, the EC directs specific attention to the favourable outputs that its AI program can achieve for European societies, how it can contribute to fulfilling the purposes of the EU as a political-administrative institution and citizens' expectations towards it.Prominently this refers to sustaining societal well-being and prosperity.This does not mean that the EC does not address AI threats.The EC does propose solutions in several of their communications, specifically in its 2021 proposal for AI regulation, the White Paper on AI or the Guidelines for Trustworthy AI.However, the EC frequently employs EI frames to redirect attention to fulfilling its historic role of technology leadership, to positive technological outcomes for EU purposes and to reassuring claims of trustworthy and human-centric AI.It seems as if the EC tries to cover up its struggle to profoundly reconcile AI and EI, as if these do not contradict, as if AI can be made to serve the European values, as if giant AI firm can come to serve democracy.Communicating European AI as directly beneficial for citizens' lives might have a stronger mobilizing effect than committing to values on paper.After all, considering a recent decline in the commitment to European values such as the commitment to fundamental rights, the rule of law or non-discrimination in EU Member States in the context of the wider 'crisis of democracy' and rise of populism, it is questionable whether the EC can reasonably guarantee AI in line with European values (Coeckelbergh & Metzinger, 2020).Research and scientific knowledge on how trustworthy and human-centric AI can be realised in practice to make democracy work is still work in progress (TU Dortmund, 2022).The fact that in its communication, the EC disregards the negative effects of (past) technological revolutions of capitalist worlds, framing the European version of AI as an uncontroversial source of societal well-being and prosperity, is in line with this observation.EI-frames are strategically used to direct the public's attention to the sunny sides of AI.One could question whether this is credible, unrealistically optimistic or even deliberately misguiding.
The EC's political communication of its AI program is part of broader identity politics seeking European integration, particularly after Brexit.In light of global competition for AI leadership, the EC emphasises that AI demands European unity, framing European AI as a cross-EU governance project for which Member States, public and private stakeholders need to join forces at the supranational level.This is directed towards the investment levels necessary for AI uptake but especially to Member States.Thus the alleged AI revolution is not only an opportunity for European integration and partnerships worldwide but also to expand and maintain the EU's relevance from within in times of Brexit, Euroscepticism and anti-EU populism.Because the EC communicates its AI program in light of global competition and AI threats, it is surprising that it does not distinguish a European version of AI through strategies of othering.Instead, the EC communicates externally that it seeks allyships based on shared valuessomething that could be achieved with the USA but probably not with China.This can be interpreted as confirmation that even if the EC underlines the importance of allyship on a global scale and addresses threats of the global race on AI, its communication practices are indeed inward-oriented to mobilize multiple EU levels for AI uptake and trust in European AI.EU Member States shall pool AI legislative action and coordinate research activities at the supranational EU level, maintaining and strengthening EU legislative powers and societal relevance; other governance actors, such as financial institutions, SMEs and big companies, shall trust in AI to generate important investment resources and technological advancement; and EU citizens shall trust and finally accept EU action on AI and EU political-administrative legitimacy in general.After all, the EC's political communication does not hint towards how specifically its approach to AI is uniquely European, its frames do not clarify how this European version of AI will actually look like.

Table 1 .
Coding scheme to analyse the Commission's approach to European identity following categories.

Table 2 .
Frame 'Using AI for Good'.withmassivepowerabuse,exploitation,alienation,extractivism,pollution, wars, etc.; also the AI revolution comes with a toll.Instead of addressing these historic challenges, the EC uses references from a shared history of industrial technological advancement as evidence for European success, while it presents AI leadership as a continuum of this modern history(European Commission, 2020c, 2020e, 2020f).References to European history and EU purposes are joined to reconcile AI and EI, for example by Commissioner Věra Jourová: What makes AI special is that it can improve all sectors of our economy and our everyday life, just like electricity or automation did (European Commission, 2019d, p. 3).We call this facet of the 'Using AI for Good' frame the 'Best of both Worlds' frame (see Table2).The EC provides an image of the best of both worlds: economic success and technological innovation that leads to global leadership and market competitiveness on the one hand, and a human-centric, trustworthy European version of AI, that first and foremost serves the Europeans, provides safety and security, resilience and democracy, on the other hand(European Commission, 2019b, 2020g, 2021a).Interestingly, the EC narrates a necessity for a European version of AI which is rooted in modern European history: 'Historically, industrial revolutions have spread from Europe to the rest of the world' (European Commission, 2018c, p. 1), as Commissioner Mariya Gabriel remarked in 2018.Thereby, regardless of European values or purposes, the EC frames European technological leadership as natural law and, thus, despite AI ambiguity, European AI's uptake as non-negotiable. came