AI in the workplace and Employment Tribunal litigation

Generative AI can help organise information and improve drafting, but may also produce lengthy, repetitive or inaccurate material.

Employers are receiving increasingly detailed complaints and data subject access requests (DSARs), while courts and tribunals are encountering AI-assisted documents and correspondence. Employers should focus not on AI itself, but on the substance and reliability of the material, and apply existing processes consistently.

AI and workplace relations

Grievances
Employees are submitting more detailed grievances containing extensive factual allegations, statutory references and requests for investigation, creating resourcing and case-management challenges for HR and legal teams.

Whilst AI may change a complaint’s presentation, it does not change the employer’s obligations. An AI-assisted grievance should still be handled considering the employer’s procedure and the ACAS Code.

The challenge is identifying the genuine complaint. A structured grievance form or initial scoping meeting can clarify the nature of the complaint and the outcome sought. The aim is to agree the issues requiring investigation, not challenge the employee’s use of AI.

Whistleblowing
AI makes it easier to frame concerns in statutory whistleblowing language. Employers should promptly identify and assess matters that may qualify for protection, including regulatory breaches, health and safety concerns and financial misconduct.

However, labelling a complaint a “protected disclosure” does not determine its legal status. Employers should examine what was raised, how it was communicated and addressed, and whether any alleged detriment followed, focusing on the facts rather than legal labels used by a worker.

DSARs
AI can facilitate broad DSARs spanning multiple systems and time periods. They cannot be ignored, but early engagement can control scope and cost. Employers should:

• engage legal advisers at an early stage;
• seek clarification about the information the individual is most interested in receiving;
• identify and preserve potentially relevant documents early; and
• record the rationale for the searches undertaken.

Communications
Communications should remain measured and concise. Direct engagement may show that an aggressive AI-assisted communication does not reflect the employee’s position.

In De Bruyn v Sanctuary at Buderim (AB2025/973), the Australian Fair Work Commission did not criticise AI use itself but observed that the style and volume of correspondence contributed to a communication breakdown. The practical lesson is to keep disputes human and focused, using direct conversations where possible.

Employment Tribunal trends and the AI effect

Employers are encountering longer pleadings, applications and targeted requests for documents, increasing legal costs, management time and search burdens.

Interim relief: act immediately
The joint Presidential Guidance on interim relief took effect on 22 June 2026. This emergency remedy is available only in certain automatic unfair dismissal claims, most commonly where dismissal is alleged to be because of a protected disclosure. Applications have increased, due to AI alerting individuals to the remedy; the Guidance also recognises that AI may add to the volume and complexity of supporting material.

The Guidance does not object to AI use but warns that AI-generated submissions may be lengthy, irrelevant or unfocused. Litigants remain responsible for ensuring that material is concise, accurate and relevant.

An employer may only receive seven days notice of an interim relief application and should establish an immediate escalation process if an application for interim relief is received. 

AI-generated submissions: accuracy and proportionality
Courts and tribunals are alert to fabricated cases, inaccurate quotations and misstated legal principles. Because AI can sound authoritative when wrong, every AI-assisted authority, quotation and statutory proposition should be checked against a reliable primary source.

In Bakshi v James Gibb Management Limited [2026] SAC (Civ) 38, the Court criticised 11 pages of apparently software-generated argument as misleading, unfocused and overstated. Two cited cases appeared not to exist. The decision confirms that litigants remain responsible for their submissions regardless of how they prepared the material.

Similarly, in Hancox v Sutherland and others [2026] EAT 139, a litigant in person filed a 300-page ChatGPT-generated skeleton argument and asked the respondents to identify inaccuracies. The EAT found the AI-generated skeleton argument unacceptable: it breached the Practice Direction, obscured rather than illuminated the issues and had not been checked by the claimant.

Employers should verify AI-assisted material and respond strategically. Identify the central procedural and substantive issues rather than treating every proposition as equally important. 

AI and witness evidence
A witness statement must reflect the witness’s own evidence in their own words. AI may help with presentation, but material rewriting can cast doubt on whether the statement genuinely reflects the witness’s account.

In Godwin v Godwin [2026] EWHC 923 (Ch), the defendant and his wife admitted using ChatGPT for grammar, spelling and presentation but did not produce their original drafts. Unable to confirm that the statements were in their own words, the court treated their evidence with caution.

When an employer intends to lead witness evidence in a court or tribunal, it is important to:

• advise witnesses not to use AI to draft or materially refine their statements;
• retain original drafts if AI provides limited assistance and be transparent about the process;
• provide appropriate support where English is not a witness’s first language or assistance is required because of disability; and
• not use AI to generate suggested answers or cross-examination responses, as this may be criticised as witness coaching.

Authenticity of digital evidence
AI can generate or alter documents, images, audio and video. Where authenticity is disputed, preserve original files and devices, retain metadata where lawful, record the chain of custody and consider forensic advice.

Disclosure, privilege and DSARs
Prompts and outputs may contain legal advice, litigation strategy, personal data or sensitive workplace information, thereby creating disclosure, confidentiality, privilege and DSAR risks. Organisations should use approved enterprise tools, minimise inputs and warn employees to avoid entering privileged or highly sensitive information into AI unless authorised safeguards apply.

AI governance and litigation readiness
AI governance should form part of litigation readiness. Employers should consider whether their policies and procedures address:

• authorised and prohibited tools, and rules on data handling and confidentiality;
• human review, approval and verification of sources and legal authorities;
• retention of prompts, outputs and original drafts, with appropriate audit trails;
• escalation where AI materially affects an investigation or litigation document; and
• training for HR, legal teams, managers and witnesses.

Final message
AI can be a valuable support tool, but it does not replace legal judgment, factual analysis or procedural discipline. Whilst AI may change an employer’s approach to handling a complaint or litigation, it does not change the underlying legal principles.

If you would like to discuss any of the issues raised above, do not hesitate to contact our employment team. 

Related News, Insights & Events