Preparing for ‘whistleblowing inflation’ in age of AI

Since its establishment in 2020, GI&T Law Office has assisted multinational and domestic corporations in implementing and operating internal whistleblowing systems. Recently, however, the author has observed a new trend: an increasing number of whistleblower reports appear to be drafted using AI tools.

While AI can make reporting mechanisms more accessible, it also presents new challenges for compliance teams. In particular, organisations may soon face what can be described as “whistleblowing inflation” – a rise in the volume and complexity of reports without a corresponding increase in the quality of the underlying information.

From vague reports to polished complaints

Traditionally many whistleblower reports were vague and unstructured. A typical complaint might read: “My manager’s harassment has been terrible lately. He keeps verbally abusing people and everyone is struggling. Please do something about it.”

This is understandable. Most employees are not legal or compliance professionals. They are unfamiliar with organising facts, identifying legal issues or explaining concerns in a structured manner. As a result, many reports lacked detail and were difficult to assess.

Today, however, employees have access to generative AI tools. By asking an AI system to rewrite or formalise a complaint, they produce reports that appear professional and sometimes resemble documents prepared by lawyers. The problem is that sophisticated language does not necessarily mean better information. Because AI can only work with the facts the user provides, the report often contains legal terminology and polished wording but add little factual substance.

In some cases, the report is more difficult to understand because key facts are buried beneath layers of AI-generated language. For example, the above-mentioned complaint can easily be transformed into a formal report alleging potential workplace harassment, psychological harm, and concerns over workplace safety. While the language sounds persuasive, the facts remain largely unchanged.

AI-driven complaint volume strains investigations

The challenge does not end with the initial report. Employees who use AI to draft complaints frequently use AI during the investigation. Compliance personnel may therefore receive lengthy responses, detailed follow-up questions and extensive written submissions that contain little new factual information.

As a result, investigators may spend more time reviewing communications. Establishing the facts is more difficult despite the apparent sophistication of the correspondence. This trend may also increase the overall number of whistleblower reports. In the past, employees tolerated minor workplace frustrations without formally reporting them. Today, an employee can describe a workplace incident to an AI chatbot and receive a response suggesting that the conduct may violate labour laws, anti-harassment policies or regulatory guidance. The same chatbot can then generate a polished whistleblower report within minutes.

Consequently, organisations may see a growing number of cases involving borderline management conduct, performance evaluation disputes, interpersonal conflicts, and workplace communication issues. At the same time, investigators must devote more resources to reviewing and processing those reports.

This creates a risk that resources intended to identify serious misconduct – such as accounting fraud, quality control violations, conflicts of interest, bribery, corruption, or antitrust violations – will instead be consumed by lower-risk workplace disputes.

Adapting whistleblowing processes for generative AI

Organisations cannot realistically prohibit employees from using generative AI when preparing whistleblower reports, and should adapt their processes.

First, organisations should take advantage of AI-powered summarisation tools that are increasingly available in modern whistleblowing platforms. These tools can help investigators quickly identify key allegations and supporting facts.

Second, where the whistleblower is known and willing to co-operate, early interviews can often reveal the real issues. Investigators frequently discover the employee’s actual concerns differ substantially from the polished narrative in the report.

Third, companies should establish effective mechanisms to distinguish serious legal or regulatory concerns from relatively minor workplace disputes. Resources should be allocated based on risk and potential impact. Finally, organisations should streamline feedback processes.

Recognise trend, adapt investigative processes

Generative AI is changing internal whistleblowing systems in ways few organisations anticipated. The primary challenge is not simply an increase in reporting volume, but the emergence of professionally drafted complaints that often contain limited additional factual information.

Organisations that recognise this trend early and adapt their investigative processes accordingly will maintain effective speak-up cultures, while preserving resources for misconduct risks that matter most.

Kengo Nishigaki is a representative partner at GI&T Law Office in Tokyo