The FIX Trading Community has submitted ten formal recommendations to the Monetary Authority of Singapore (MAS) regarding AI risk management in financial markets. This response addresses the growing integration of large language models and machine learning within algorithmic trading. The proposed framework aims to mitigate the risk of market instability and “contagion” caused by the rapid, interconnected nature of global AI-driven trading systems.
A central pillar of the proposal is the establishment of a globally recognised definition and taxonomy for AI to prevent regulatory arbitrage. FIX suggests anchoring new guidelines to existing standards, such as MiFID RTS 6 and DORA, while making company boards directly accountable for the effectiveness of AI governance. The recommendations also advocate for cross-functional oversight of third-party suppliers and the implementation of disclosure patterns for AI data and training, similar to food labelling.
Furthermore, FIX proposes expanding risk assessments to include “change sensitivity” and “interconnectedness” to account for non-linear AI behaviour. The group suggests that AI inventories must be more granular than traditional algorithm logs and that any fine-tuning or data refreshes should undergo rigorous materiality tests. By leveraging its expertise in algorithm certification, FIX aims to ensure that AI lifecycle controls remain robust as the technology evolves.
Subscribe to our newsletter
