A global law firm partnered with Control Risks to trace misappropriated funds in a multi-faceted investigation. Using aiR Review they uncovered key evidence and connections in a complex case of financial irregularity.
AI enabled a faster, more insightful investigative process:
- First-pass review accelerated to identify key evidence immediately
- Review focus sharpened AI-powered categorization and rationales
- A multidisciplinary investigation unified by a single source of truth
Following the trail of diverted funds
A global law firm engaged Control Risks to assist with an investigation into financial irregularities within an oil and gas company headquartered in Africa.
The case centered a 2016 transaction executed by an energy company where a substantial payment, held in trust to facilitate a share repurchase, was unlawfully redirected to a third-party entity instead of the intended recipient.
Leading up to and following the transaction, numerous documents and communications were exchanged, including invoices, letters of intent and banking details. Despite formal agreements and signed instructions for payment release, the expected funds were not received by the intended party. Instead, they were transferred to an affiliated account, prompting concerns about the transaction’s execution and oversight.
Control Risks was brought in to investigate the redirection of the misappropriated funds and identify any individuals or entities potentially involved in the misappropriation.
Merging forensic expertise with AI prompts
Control Risks mobilized a multidisciplinary team comprising specialists from Discovery and Data Insights, Investigations and Forensic Accounting and Business Intelligence. The forensic consultants supported the client in collecting digital accounting records and other relevant documentation, including email and communication data from key individuals.
Once the data was processed and made accessible, project managers worked closely with investigative and forensic accounting experts to develop targeted search terms and organize review batches for analysis. Meanwhile, the Business Intelligence team conducted thorough due diligence on individuals and third parties identified during the investigation, uncovering connections such as family relationships and shared business interests.
Initial keyword searches surfaced approximately 10,000 responsive documents. A sample of one hundred manually reviewed documents, including key items provided by legal counsel, was used to refine an AI prompt. Once finalized, the prompt was applied to the full set of responsive documents, allowing the AI to categorize each document by relevance and enabling a focused and efficient first level review.
Intelligent categorization
Control Risks applied the AI model across the entire set of 10,000 documents, which were categorized as “Very Relevant”, “Relevant”, “Borderline Relevant”, or “Not Relevant”.
The investigations team then concentrated its second-level review on documents identified as “Very Relevant”, “Relevant”, or “Borderline Relevant”. Our internal investigations team acted as AI interpreters, reviewing and relying on the AI prediction rationale and considerations, the latter providing a detailed counterargument to consider when reviewing the AI prediction.
To maintain quality assurance, a 10% sample of the documents classified as “Not Relevant” was also reviewed. Additional key documents were identified within the “Very Relevant” and “Relevant” categories and were used to support the final report provided to legal counsel.
Actionable intelligence delivered with speed and precision
The fusion of AI and expert analysis allowed Control Risks to deliver a decisive outcome for the client. This approach enabled a targeted review, reduced the volume of manual analysis required, and strengthened the overall findings of the investigation.