UK Trends and Developments Contributed by: Steven Cochrane, Hannah Netherton, Anna Cope and Gillian MacLellan, CMS
AI in Investigations: Risks, Uses and Guardrails AI is beginning to reshape workplace investigations on both sides of the table. Employees are increasingly using AI tools to draft grievances or whistle-blowing reports, generate data subject access requests and prepare pleadings for employment-related claims. This has led to greater volume, complexity and speed in communications, and uneven quality. The use of AI by employees raises various novel risks in the context of workplace investigations, and investigation proce - dures should be adapted to mitigate those risks. First, confidentiality and data protection. Employees uploading internal emails, interview notes, chat logs or personnel data to open-source AI tools is a serious risk and can lead to the loss of the business’ con - fidential information and data breaches. Acceptable Use policies, whistle-blowing, grievance and discipli - nary procedures should make it explicit that company confidential information and personal data must not be uploaded to open-source AI platforms. This mes - sage should be reinforced through training, and con - sequences for misuse followed through. Second, accuracy. AI‑generated submissions may appear polished but can often be wrong. Examples include non‑existent cases, made-up quotes, and speculative or untrue allegations presented as fact. Investigators should be alert to the potential for “hal - lucinations” and seek to challenge dubious references and information. Employees should also be required, through relevant policies and procedures, to take ownership (and accountability) for information pro - vided as part of a workplace investigation.
Third, process integrity and volume control. AI enables rapid‑fire correspondence and “weaponised” com - plaints which require careful and sensitive manage - ment by an investigator. It is important for investiga - tors to avoid becoming overwhelmed by information and to maintain proportionality in an investigation. This can be helped by timely acknowledgement of communications, consolidating issues, setting clear expectations around the nature and scope of the investigation and setting reasonable timelines. For businesses, AI can be a valuable tool in man - aging workplace investigations in an efficient and proportionate way. Use cases include summarising evidence, preparing and verifying chronologies, and automated notetaking at witness interviews. Guard - rails are essential – human oversight and an under - standing of the limitations of AI will be important to mitigate the legal, regulatory and reputational risks of inappropriate AI use and will help to ensure that an investigation holds up to internal and external scrutiny.
416 CHAMBERS.COM
Powered by FlippingBook