Bloomberg has expanded Bloomberg Vault with two artificial intelligence models designed to identify electronic communications that may point to insider dealing or breaches of employee personal trading policies. The additions extend the platform’s surveillance coverage to two risk areas that can require compliance teams to examine communications alongside trading and account activity.
The Insider Dealing AI Policy looks for communications that may indicate the improper acquisition, disclosure or use of material non-public information (MNPI). The Personal Trading AI Policy targets communications concerning employees’ trading activity and investment accounts, including possible failures to comply with firms’ internal personal trading requirements.
The models join Bloomberg Vault’s existing suite of AI-powered surveillance policies, which covers market conduct, non-market conduct and conflicts of interest. Bloomberg said the expanded suite is intended to help compliance teams identify potential risks within electronic communications and concentrate their reviews on more relevant alerts.
Each model has been built for a defined risk scenario and is supported by documentation explaining its construction, risk focus and intended behaviour. According to Bloomberg, large language models are applied following model inference to improve the relevance of alerts and reduce false positives.
The documentation is also intended to help firms assess how the models operate within their AI governance frameworks. This is particularly relevant where compliance teams must explain the basis for surveillance alerts and evaluate externally supplied AI systems before deploying them in regulated workflows.
“Bloomberg’s AI-powered surveillance models have substantially improved alert quality and reduced false positives,” said Jotham Banyikidde, Investment Compliance Senior Associate at Impax Asset Management. “This allows our team to focus on meaningful review and oversight. Just as importantly, the models are transparent and well documented, which supports our internal AI impact assessment and gives us confidence in how alerts are generated.”
The release follows Bloomberg’s introduction of BSpeech, its AI-powered, multilingual voice transcription service. Bloomberg said the combination extends Vault’s AI capabilities across electronic and voice communications, allowing firms to manage surveillance and communications governance across multiple channels through a single workflow.
Subscribe to our newsletter
