When AI Systems Mislead. Rethinking Market Integrity in the Age of AI

Artificial intelligence systems can spread false or misleading information, thereby endangering the informational efficiency and integrity of financial markets and eroding public confidence. This conflicts with the objective of the European legislator to promote full and proper market transparency It is thus essential to evaluate the adequacy and effectiveness of current enforcement mechanisms in addressing the challenges posed by emerging technologies within financial markets. ‘Strong’ AI systems—those equipped with self-learning capabilities—can generate autonomous and unpredictable outputs that diverge significantly from the inputs provided by their manufacturers, programmers, or users. These characteristics raise substantial difficulties in applying existing civil, administrative, and criminal liability regimes. In particular, they may preclude the identification of intent or negligence, while also complicating the establishment of a causal link between a human agent’s conduct and the harm in question. A potential solution to the problem of unpunished dissemination of false or misleading information lies in an approach already employed in the context of digital platforms. Specifically, negligence may be attributed to the user of a ‘strong’ AI system who fails to act to remove or rectify demonstrably false information once known, or where such a system—due to an unresolved defect—repeats the dissemination of similar misinformation. Finally, additional enforcement strategies can be found in the powers granted to trading venues and competent supervisory authorities, which together may help safeguard the legislative aim of promoting transparency and integrity in financial markets.