Global Relay, the leading provider of electronic communication compliance and archiving solutions, released its State of AI in Surveillance Report 2025, discovering that 31% of industry respondents have already or intend to implement AI in surveillance processes in the next 12 months, as financial firms come under increased regulatory pressure. The report also revealed a 19% reduction in the number of firms reluctant to implement AI since June 2024, suggesting that attitudes to AI are changing for the positive.
The global survey examines how compliance, surveillance and risk leaders in financial services are integrating AI into compliance workflows, charting benefits, challenges and regulatory implications of using the technology. With over $8 billion in SEC fines levied last year for compliance failures, regulators are cracking down on outdated lexicon-based monitoring tools.
Lexicon-based approaches require surveillance teams to manually define all the keywords for which their system needs to watch. This means it is difficult to capture every term or phrase variation that personnel may use relating to high-risk areas – let alone identify misconduct attempts that don’t fit into prescribed lexicon buckets. Implementing AI-enabled surveillance can reduce false positives (the top goal of 23% of respondents), improve risk identification (20%), execute voice transcription (14%) and more.
Donald McElligott, VP of Compliance Supervision, says: “The whole world of surveillance has been about dealing with overwhelming volumes of false positives. Everything you did was eliciting 100 false positives for every one legitimate red flag to investigate. AI flips this over.” He continues: “The days of throwing people (and thus, salaries) at a big pile of data and sifting through to find a couple of needles in a haystack are gone. AI only identifies real risks, making compliance more efficient and precision-driven.”
The research found that data security remains the biggest barrier for firms adopting AI into surveillance workflows (25%), followed by budget (21%), explainability and transparency (17%) and pushback from internal teams (13%).
Rob Mason, Director of Regulatory Intelligence, comments: “Explainability has long been a barrier to AI adoption – the fear that, in the event of regulatory investigations, firms using AI-driven processes will be unable to explain to the regulator how a decision was made. However, with AI outpacing lexicon-based approaches, it begs the question as to whether regulators should be considering how they can evaluate this new technology. The benefits are overwhelming.”
The research also indicates that 38% of respondents are watching how the industry develops to determine whether to use AI in surveillance processes, as regulators across jurisdictions have taken several different approaches to AI’s adoption.
For more information on State of AI in Surveillance Report 2025, please see here.
