Tessa Sophie Hoffmann & Chrisa Alexiou (University of Groningen)

Photo credit:  Jack Sexton, via Wikimedia commons

Introduction

As Tony Milligan wisely observed in 2011, love is a complex emotion, capable of bringing immense joy and unexpected delight, but also capable of causing pain and disappointment. While the pursuit of love has existed long before the digital age, technology and artificial intelligence (AI) have dramatically changed the landscape, with over 48 million Europeans turning to dating platforms to find potential partners. This surge in online dating has created a lucrative market, with industry giants like Tinder, Grindr, and OKCupid offering user-friendly platforms that appear deceptively straightforward. However, behind the seemingly simple interfaces lie complex and often unregulated issues.

These platforms rely heavily on algorithms to filter potential matches based on factors like location and age preferences. Tinder, for instance, is known for its geolocation-based swiping system, where users make quick judgments about potential matches. This use of algorithms, particularly when personal information is involved, raises concerns about transparency, privacy, data protection, liability for harm, bias, and discrimination. Existing regulations, such as the E-Commerce Directive, struggle to adequately address the intricacies of these algorithmic systems, highlighting the need for updated legislation specifically tailored to online dating platforms.

The Role of Algorithms and AI in Dating Applications

Location-based services (LBS) are fundamental to online dating apps, enabling them to track user locations with consent. This creates a bridge between the physical and digital worlds, allowing these platforms to connect individuals based on proximity.

The integration of AI takes online dating a step further. Unlike simple algorithms, AI utilizes machine learning to analyze vast amounts of user data, identifying patterns and preferences. This allows AI to suggest matches based on a user’s behavior and interactions, potentially leading to more compatible pairings.

Different dating apps employ different algorithmic approaches. Tinder, for example, uses an Elo rating system, similar to that used in chess, to rank users based on their “desirability,” as reflected in how others interact with their profiles. OKCupid, on the other hand, asks users a series of questions about their beliefs and values, claiming to use this information to create more meaningful matches. However, the effectiveness of such algorithms and the influence of paid features, such as profile boosts and “super likes,” remain debatable.

Additionally, dating platforms utilize AI to monitor user content, flagging inappropriate or suspicious behavior. This includes identifying fake profiles, scams, and spam. However, the line between protecting users and prioritizing engagement can become blurred, raising concerns about the extent of control these platforms exert.

Consequences and Risks of Algorithmic Love

Lack of Algorithmic Transparency

While efforts have been made to regulate online platforms through legislation like the Proposed DSA, the Unfair Commercial Practices Directive, and the Artificial Intelligence Act, the unique challenges posed by dating apps necessitate a more targeted approach. For instance, while the Proposed DSA mandates platforms to remove illegal content and report criminal activity, the reliance on algorithms to flag such content can lead to the unjust suspension of users due to the algorithm’s limited understanding of context and nuance. Furthermore, proving algorithmic liability for failing to identify harmful content remains a significant challenge.

The lack of transparency surrounding the algorithms used by dating platforms exacerbates these concerns. Companies are often reluctant to disclose how their algorithms work, citing competitive reasons. This lack of transparency prevents users from understanding how they are being categorized and matched, potentially leading to ethical dilemmas.

Privacy and Data Protection

Dating apps collect a significant amount of personal information, often more so than other social media platforms. The use of geolocation services and GPS tracking raises serious privacy concerns, as does the potential for data breaches and misuse. While the EU’s General Data Protection Regulation (GDPR) provides some safeguards regarding data protection and user rights, its effectiveness in the context of AI-driven dating apps is questioned. Legal experts argue that the GDPR’s provisions on transparency, data minimization, and automated decision-making are insufficient to address the complexities of AI. The opacity of these systems makes it difficult for users to understand how their data is being used and to hold platform providers accountable.

Liability for Damage/Legal Personhood Issues

The issue of liability for harm caused by AI systems on dating platforms presents a significant legal challenge. As AI lacks legal personhood, holding it directly accountable for damages is problematic. While the EU Parliament has debated granting AI legal personhood, the current stance leans towards ensuring human accountability for AI systems.

The EU Commission, in its reports on AI liability, acknowledges the need to protect users from harm caused by emerging technologies. However, the complexity, self-learning capabilities, opacity, and vulnerability to cybersecurity threats associated with AI make it difficult to establish liability and ensure compensation for victims. The Proposed AI Liability Directive seeks to address this by harmonizing liability rules across the EU and shifting towards a strict liability approach for AI systems, potentially making it easier for victims to seek redress.

Bias and Discrimination

The algorithmic ranking systems used by dating apps often perpetuate existing societal biases, particularly concerning physical attractiveness, gender, sexuality, and age. The emphasis on visual appeal and the “swipe left or right” mechanism can lead to superficial judgments and reinforce harmful stereotypes.

The EU Agency for Fundamental Rights, in a 2018 report, highlighted the potential for algorithmic discrimination. The EU Charter’s Article 21 prohibits discrimination, and the European Parliament has urged both the Commission and Member States to address algorithmic bias and promote transparency in data processing. Paid features that allow users to boost their profiles further exacerbate this issue, creating a system where access to potential matches is influenced by financial means.

While traditional recommendation systems used by dating apps can perpetuate these biases, researchers suggest that reciprocal recommendation systems, which consider the preferences of both parties, could lead to fairer and more satisfying matches.

The Legal Framework: A Call for Reform

The significant power wielded by dating platforms and their impact on users’ rights necessitate a tailored regulatory framework that goes beyond existing legislation. The current lack of comprehensive regulation in the EU concerning AI systems, as acknowledged in the White Paper on AI, risks undermining fundamental rights and non-discrimination principles.

The European Commission’s risk-based approach to AI regulation, as outlined in the Proposed AI Act, subjects high-risk AI systems to stricter obligations. Article 6 of the Act defines high-risk systems as those that impact safety and require third-party conformity assessments. Given the potential for harm, dating platforms with complex algorithms could be categorized as high-risk and subject to greater scrutiny. Additionally, Article 5 of the Proposed Act prohibits AI systems that manipulate behavior, exploit vulnerabilities, or discriminate, aiming to protect users from psychological harm.

The Act also proposes regulatory sandboxes as a means for developers to test and refine AI systems in a controlled environment, potentially leading to safer and more user-centric designs for dating platforms.

Addressing liability concerns arising from AI-driven harm is crucial for building trust in these systems. The Proposed AI Liability Directive seeks to overcome the challenges of proving fault in cases involving AI by introducing a strict liability regime and harmonizing rules across the EU. This will provide greater clarity and legal certainty for both users and providers.

Conclusion

The increasing prevalence of online dating and the rapid advancement of AI technologies necessitate a more focused and comprehensive regulatory framework. While the EU has taken steps to address the broader challenges of AI, specific regulations tailored to online dating platforms are essential to ensure user protection, fairness, and accountability.

Existing EU initiatives like the Proposed AI Act and the AI Liability Directive provide a solid foundation for regulating dating apps. However, continuous improvement and adaptation are crucial to keep pace with technological advancements. Integrating human oversight into algorithmic processes, particularly in sensitive areas like online dating, could further enhance fairness and prevent unintended harm.

Ultimately, a balanced approach that fosters innovation while safeguarding user rights is crucial. Harmonized EU-wide regulations are essential to avoid a fragmented legal landscape and ensure that all European citizens benefit from a safe and trustworthy online dating environment.