Taylor Wessing Slovakia

A successful first Global Dialogue on AI Governance would deliver practical, operational outcomes reflecting that AI is not only a driver of innovation, but also a force multiplier for fraud—enabling scalable schemes such as deepfake impersonation, automated social engineering, and synthetic identity fraud. First, success would mean a shared understanding among governments, regulators, law enforcement, and the private sector that AI-enabled fraud is a systemic risk requiring coordinated, cross-border responses. Traditional prevention frameworks are no longer sufficient when adversaries can operate at scale with high credibility. Second, the Dialogue should lead to concrete commitments to strengthen prevention and detection through AI. This includes promoting responsible use of AI by financial institutions and corporations to detect anomalies, identify deepfakes, and flag high-risk transactions in real time, alongside baseline standards for AI-assisted fraud monitoring. Third, success requires improved mechanisms for rapid information-sharing and cooperation. AI-driven fraud operates across jurisdictions within hours, making faster data exchange, stronger FIU collaboration, and effective public–private partnerships essential. Finally, the Dialogue should identify clear pathways for capacity-building, equipping law enforcement and judicial authorities with the tools and expertise needed to address AI-enabled crime. In sum, success lies in moving from principles to actionable, cooperative solutions that use AI as part of the response to AI powered crimes.

Safe, secure and trustworthy AI;

It reflects the urgent need to address AI as a tool for scaling fraud. AI lowers barriers to sophisticated schemes, enabling deepfake impersonations, automated phishing, and synthetic identity fraud at unprecedented speed and scale, creating systemic risks for businesses, financial systems, and public trust. Ensuring AI is safe and trustworthy must include strong safeguards against misuse. This involves reducing exploitability of AI systems, improving transparency of AI-generated content, and developing reliable tools to detect manipulated audio, video, and text. At the same time, defenders must be able to use AI effectively. Financial institutions, companies, and law enforcement need AI tools to detect anomalies, identify fraud patterns, and respond in real time, otherwise the gap between attackers and defenders will widen. Trust also depends on effective cooperation. Addressing AI-enabled fraud requires coordinated action between the private sector, regulators, and law enforcement, including faster information-sharing and aligned risk mitigation standards. Prioritising safe, secure and trustworthy AI is therefore essential to ensure AI strengthens, rather than undermines, trust in digital economies and institutions.

A key cross-cutting issue that is not sufficiently emphasised is the role of AI as a force multiplier for fraud, which is already the most financially damaging category of crime globally. Fraud is no longer a marginal or isolated phenomenon; it represents a systemic economic threat, and AI risks amplifying it to an unprecedented scale. If clear boundaries and safeguards are not established early, AI will enable criminals to industrialise fraud-automating deception, increasing credibility through deepfakes, and targeting victims at scale with minimal cost. The resulting social harm would be significant, eroding trust in digital communication, financial systems, and even public institutions. This could materially offset the positive benefits that AI is expected to bring. For this reason, the potential misuse of AI in criminal activity should be treated as a priority from the outset of AI governance discussions, not as a secondary or downstream concern. In parallel, strong emphasis should be placed on the use of AI as a defensive tool. AI-driven solutions for detecting AI-generated fraud, identifying anomalies, and enabling rapid response are essential to maintaining balance between attackers and defenders. Importantly, failure to address AI-enabled fraud early carries regulatory consequences. If AI-driven harm escalates significantly, it will likely trigger reactive, and potentially overbroad, regulatory interventions and restrictive government measures. This could hinder innovation and reduce the overall societal value of AI. Addressing AI-enabled fraud proactively is therefore not only a matter of crime prevention, but also a prerequisite for sustainable and proportionate AI governance.

Governance gaps in ensuring safe, secure and trustworthy AI are already affecting our region and sector, particularly through the rapid rise of AI-enabled fraud. Fraud is the most financially damaging form of crime today, and AI is significantly amplifying its scale, speed, and sophistication. Deepfake impersonations, automated social engineering, and synthetic identity fraud are increasingly targeting businesses and financial institutions, exposing weaknesses in existing prevention and response frameworks. The most significant challenge is that regulatory and institutional responses are not keeping pace with technological developments. There is still insufficient focus on the misuse of AI in crime, limited cross-border coordination, and uneven capacity among law enforcement and financial institutions to detect and respond to AI-driven threats. This creates a growing asymmetry in favour of attackers. At the same time, there is a clear opportunity to leverage AI as part of the solution. AI can enhance fraud detection, enable real-time monitoring, and improve cooperation between private sector actors and authorities. Strengthening information-sharing mechanisms and embedding safeguards into AI systems from the outset can significantly reduce risks. Addressing these gaps early is critical. If AI-driven fraud continues to scale unchecked, it will cause substantial social and economic harm and likely trigger overregulation, potentially undermining the broader benefits of AI.

It can play a critical role as a platform for aligning stakeholders with fundamentally different perspectives, incentives, and risk exposures. AI is inherently dual-use: the same tools that drive innovation and efficiency can also be exploited for fraud and other forms of harm. Effective governance therefore depends on a shared understanding of both the opportunities and the risks. A key contribution of the Dialogue is to facilitate structured exchange between governments, the private sector, financial institutions, technology providers, and law enforcement. Each of these actors approaches AI from a different standpoint-whether innovation, risk management, security, or enforcement-and their concerns are often shaped by their respective mandates and business models. Without mutual understanding, regulatory fragmentation and ineffective responses are likely. The Dialogue can help bridge these gaps by promoting transparency around how AI systems are developed, deployed, and misused, and by encouraging stakeholders to articulate their constraints and expectations. This is particularly important in addressing AI-enabled fraud, where effective prevention and response depend on close cooperation, timely information-sharing, and aligned incentives. In addition, the Dialogue can support the development of common principles and practical cooperation frameworks, including public-private partnerships, cross-border coordination mechanisms, and shared standards for risk mitigation. Ultimately, its value lies in creating a trusted environment where stakeholders can move beyond abstract principles and work towards coordinated, operational solutions that reflect the dual-use nature of AI and balance innovation with protection.

While I am not aware of a dedicated global initiative specifically addressing AI-enabled fraud, a strong model that the ai dialogue could build upon is the World Economic Forum's "Gatekeepers: A Unifying Framework" (2021), developed together with the World Bank and other partners. This framework is designed as a value-based self-regulatory model for private-sector intermediaries ("gatekeepers") - such as banks, lawyers, and corporate service providers - who are strategically positioned to prevent or interrupt illicit financial flows. Importantly, it recognises that private actors are often best placed to detect and prevent misuse in real time and promotes coordinated, cross-industry action. A similar "gatekeeper" approach could be highly relevant for AI governance. Developers of AI systems, platforms, financial institutions, and other deployers of AI could assume responsibility for embedding safeguards, monitoring misuse (including fraud), and cooperating across sectors and jurisdictions. This is particularly critical given the dual-use nature of AI and its capacity to scale fraud globally. The added value of the ai dialogue would be to elevate such a model to the global level - aligning expectations across jurisdictions, fostering trust among stakeholders, and encouraging voluntary but structured commitments. By promoting proactive self-regulation, the ai dialogue can help demonstrate that risks - especially large-scale AI-enabled fraud—are being effectively managed. This is essential to avoid fragmented or overly restrictive regulatory responses. If stakeholders act early and responsibly, governance can remain balanced, supporting innovation while mitigating systemic risks.

Different stakeholders should contribute to the AI Dialogue through a structured, multi-stakeholder format that reflects their distinct roles and incentives. Building on the "gatekeepers" approach, technology providers, financial institutions, and other key actors should take responsibility for embedding safeguards, monitoring misuse, and sharing insights on risks such as AI-enabled fraud. Governments and regulators should provide clear expectations and enable cross-border cooperation, while law enforcement contributes operational experience. It should combine policy discussions with practical, case-based exchanges and promote voluntary, principle-based commitments (integrity, transparency, accountability). This structure would support proactive self-regulation and help prevent fragmented or overly restrictive regulatory responses.

Law enforcement and victims' organisations are underrepresented despite direct exposure to AI-enabled fraud. Law enforcement offers operational insight into emerging threats, while victims highlight real-world harm. They should be included through formal participation, dedicated tracks on misuse, and case-based discussions to ensure AI governance reflects both risks and societal impact.

A dual-track format combining a public platform for policy dialogue with a private, trusted forum for sensitive exchanges would be effective. Stakeholders should commit to self-regulation against AI misuse, supported by case-based simulations and practical cooperation. This approach fosters trust, enables real-time learning, and reduces the need for reactive overregulation.

A relevant example is the World Economic Forum's "Gatekeepers: A Unifying Framework" (2021), which I contributed to. It promotes value-based self-regulation (integrity, transparency, accountability) by private-sector actors to prevent misuse. A similar approach in AI governance can embed safeguards, strengthen cooperation, and proactively mitigate risks such as AI-enabled fraud while avoiding overregulation.