Key Points

  1. AI-powered threats are becoming increasingly sophisticated, putting companies at risk of financial fraud.
  2. Companies must reassess their fraud defenses and map their transactions to identify vulnerabilities.
  3. The key to preventing financial fraud is to identify and address the seams between systems where transactions are most exposed and least owned.

The Risks of AI-Driven Financial Fraud

The article discusses the growing threat of AI-powered financial fraud, which is becoming increasingly sophisticated. Companies must reassess their fraud defenses and map their transactions to identify vulnerabilities. The key to preventing financial fraud is to identify and address the seams between systems where transactions are most exposed and least owned.

The article cites two examples of AI-driven financial fraud: the 1960s vegetable oil fraud and the 2025 Arup case. In both cases, the perpetrators exploited the trust in a single piece of information, such as a warehouse receipt or a familiar face on a video call.

The article argues that companies have become very good at defending what they own, but the next decade of fraud risk will turn on how well they defend what sits in between. An org chart shows who owns which system, but the transaction shows where the risk travels, and that is the map criminals are working from when they try to take advantage of companies' weaknesses.

At a Glance

ArupBritish engineering firm
victim of AI-driven financial fraud
Allied Crude Vegetable Oil RefiningNew Jersey commodities trader
perpetrator of 1960s vegetable oil fraud
FBI$20.9 billion
reported losses due to business email compromise in 2025

Why Companies Are Vulnerable to AI-Driven Threats

Why Companies Are Vulnerable to AI-Driven Threats

Companies rely heavily on informal controls, such as familiar signals, that have evolved over time. However, these signals are no longer reliable due to the availability of AI tools that can counterfeit them.

The article argues that companies must map their transactions to identify vulnerabilities. This involves tracing each transaction from origin to final accounting entry, across functions and legal entities. The seams between systems are where transactions are most exposed and least owned, and it is here that companies must focus their defenses.

Preparing for the Next Decade of Financial Fraud

The article suggests that companies should identify the seams between systems where money leaves and cannot be recalled. They should then put in place additional checks to verify the evidence received at these seams. This may involve naming a person to oversee each seam and giving them the authority to stop payments that cannot be verified.

Preparing for the Next Decade of Financial Fraud

Preparing for the Next Decade of Financial Fraud

Companies should identify the seams between systems where money leaves and cannot be recalled. They should then put in place additional checks to verify the evidence received at these seams. This may involve naming a person to oversee each seam and giving them the authority to stop payments that cannot be verified.

Companies should also consider implementing approval thresholds, requiring more signoffs as the dollar amount rises. This can help to add friction to the payment process and reduce the risk of financial fraud.

Risk & Opportunity Assessment

Commercial RiskMediumCompanies are vulnerable to AI-driven financial fraud due to the availability of sophisticated AI tools.
Competitive RiskLowCompanies are not yet competing with each other in terms of AI-driven financial fraud defenses.
Regulatory RiskLowRegulatory requirements for financial fraud defenses are not yet clear.
Reputation RiskMediumCompanies that fail to defend against AI-driven financial fraud may suffer reputational damage.
Technology DisruptionHighAI tools are becoming increasingly sophisticated and are disrupting traditional financial fraud defenses.
Commercial OpportunityMediumCompanies that develop effective AI-driven financial fraud defenses may be able to gain a competitive advantage.