SmarshCONNECT 2023: The Year of Agile Compliance

We’ve also seen how organizations are adapting and refining the ways they're doing conduct surveillance to address the explosion of new content sources. The ability to keep up with both the variety and volume of digital communications in today’s hybrid work environments requires shifting the mindset from using multiple single-purpose applications to a modern, scalable and well-maintained infrastructure.

Having seamless surveillance workflows is simply impossible with disparate systems. However, a compliance strategy that's powered by a single, powerful system enables firms to mitigate organization-wide risks more effectively while producing previously untapped insights.

As highlighted throughout the day by our customers, these insights are needed not only to help spot the latest forms of risk, but to also serve as a true, integrated source of intelligence that can drive innovation in an AI-first world.

Our break-out discussions later in the afternoon further explored the topic of surveillance. We found that the majority of attendees shared the common goal of expanding AI-based risk coverage to address the growing variety of social, mobile and collaborative content. Our partners at FTI Consulting mentioned that they’ve seen a 400% increase in Microsoft Teams data, as well as the increased presence of non-text characters such as emojis in investigative workloads.

Regarding the hype versus reality of AI/machine learning (ML), our panelists emphasized the importance of understanding how AI is adding value and addressing a specific use case. In Smarsh’s experience, when ML is deployed, firms have seen dramatic reductions in false positive rates and more escalations, enabling 100% review vs. review of samples only.

We expect it to follow the same journey as other emerging technologies, such as predictive coding, which has already gained mass adoption. What will remain critical is to preserve agility to respond to changes in digital communications, which is dependent on processes and training more than the underlying AI technology itself.

As expected, almost every discussion ultimately turned to ChatGPT and other flavors of large language models (LLM) and reinforcement learning. While the outcomes of ChatGPT include efficiencies, reducing alerts, and lowering costs, risk assessments will continue to depend on human judgment. It’s not a scary autonomous machine that will replace humans, but it can serve as an additive tool.