Artificial Intelligence and the Market Abuse Regulation | Professor Filippo Annunziata

Published (in revised form) on 29th January 2025 in Journal of Financial Supervisors Academy Volume I.

When Malta created a framework for Virtual Financial Assets in 2018, it set a precedent for how regulators can engage with technological disruption ahead of the curve. A new analysis by Professor Filippo Annunziata, published in Volume I of the Journal of Financial Supervisors Academy, examines a parallel challenge now confronting Maltese-listed issuers, investment firms and the wider EU capital markets community: how Artificial Intelligence (AI) is reshaping the rules on inside information and market manipulation under the Market Abuse Regulation (MAR).

Annunziata sets out how AI can play a constructive role in helping issuers comply with their disclosure obligations under Article 17 of MAR, which requires inside information to be disclosed to the public as soon as possible. With their capacity to process large volumes of data and to learn over time, AI systems can help identify potential inside information at an early stage, monitor its evolution through prolonged corporate processes and support assessments of likely market impact. For larger Maltese issuers and groups with complex internal structures, such tools could materially strengthen the timeliness and accuracy of disclosure.

The analysis is more cautious on the question of full automation. While the technology is increasingly capable of drafting and transmitting market announcements, Annunziata argues that the decision to disclose inside information, or to delay disclosure under MAR's restrictive conditions, should remain firmly with the issuer's management body. National civil and administrative liability regimes are not yet aligned to allocate responsibility cleanly when an autonomous system causes a breach, and human oversight is identified as essential at the final decision point.

On market manipulation, Annunziata reviews the existing MiFID II framework for algorithmic and high-frequency trading and concludes that it remains broadly fit for purpose. A more novel suggestion is the embedding of manipulation-prevention safeguards directly within trading algorithms, requiring firms to build in ex-ante blocking mechanisms that prevent an algorithm from operating where a reasonable risk of manipulation is identified. The approach would complement the risk-based logic of the EU Artificial Intelligence Act, which differentiates obligations according to the level of risk a system poses.

For Malta, the analysis offers a constructive path forward. With the MFSA already supervising issuers and investment firms operating under MAR and MiFID II, integrating AI into compliance frameworks presents an opportunity to strengthen market integrity, provided the rules keep humans firmly in the loop on the most consequential decisions.

Read the full article here.

This summary is part of a weekly series curated by Business News Malta, showcasing articles from the Journal of Financial Supervisors Academy (JFSA) Volume I, published September 2025.