Knowledge

The AI revolution has infiltrated almost every facet of modern life, and the domain of people management has been no exception. From the perspective of HR professionals, the undoubted benefits of effective use of AI systems in automating tasks, conducting research and summarising documents must be set against the risks of AI hallucinations, failure to apply critical thinking and over-reliance on AI tools.

Where an employee presents AI-generated work as their own, responsibility for any resulting harm may not rest with the employee alone. Although reliance on false information may be negligent regardless of source, the position may differ where the employer has encouraged the use of AI or supplied the relevant software. The issue becomes more complex where a manager has reviewed and approved the work, raising questions of shared responsibility and supervisory accountability.

Employers should therefore treat AI use as a governance issue as well as a productivity tool. Clear policies, human oversight and appropriate training remain essential.

See our separate briefing on AI use in grievances, disciplinary matters and tribunal claims - guidance for HR professionals

Recruitment

The use of specialised software to collect, sort, and organise job applications has been standard practice for many years. AI tools can now go further by screening candidates, generating or refining interview questions and assessing applications against role criteria, including whether an employee will be a good cultural fit.

AI can make recruitment more efficient, but it also creates legal risk. Poorly configured tools may produce indirectly discriminatory outcomes if the criteria used correlate with protected characteristics or systematically exclude suitable candidates.

For example, an AI tool trained to prioritise candidates based in the Channel Islands may disadvantage suitable candidates who are indeed resident in the jurisdiction, but whose relevant experience was gained elsewhere, by placing too much weighting on where they have worked or trained.

Regular bias audits should be carried out on AI hiring tools to guard against potentially discriminatory outcomes. This may require human intervention, but a little traditional oversight may go some way to guard against the unintended consequences of automated decision-making.   Organisations should also ensure that candidates receive clear privacy information in advance. If automated decision-making forms part of the recruitment process, candidates should be told how their data will be used and what role automated processing will play, including any right to request human intervention.

Data protection

There are potentially serious implications to using AI systems to process confidential data. The key consideration in relation to data protection is what AI platform you are using - is it an internally controlled enterprise system, or a public-facing third-party platform? The latter is more likely to create confidentiality and data breach risk if used without appropriate control.

Since the wider adoption of generative AI tools, there has been a particular focus on the impact the use of AI platforms may have on processing of personal data. Organisations have a legal duty to ensure that employees acting on their behalf process personal data lawfully, fairly and transparently when using those tools.

The Guernsey Office of the Data Protection Authority recently published "Ten-step practical AI guidance" to help organisations use AI responsibly where personal data is involved. In summary, organisations should:

• check if you are processing personal data;
• understand your role and lawful processing condition;
• carry out a data protection impact assessment;
• be transparent, fair and accountable;
• handle training data responsibly;
• respect individuals' rights;
• manage risks – do not aim for zero-risk;
• secure the data and the AI system;
• keep records and be ready to explain; and
• maintain ongoing oversight.

The Jersey Office of the Information Commissioner has also identified artificial intelligence as one of its strategic priorities for 2026–2028.

Hallucinations, accountability and supervision

AI-generated content can improve efficiency, help standardise documents and support drafting. The main risks arise where users rely on inaccurate output, fail to verify authorities or facts, or treat the tool as a substitute for judgement.

For obvious reasons, when things go wrong using AI, businesses do not publicise it. Recent court cases in England and Wales, Guernsey and Jersey show a consistent approach to AI - the responsibility for checking AI-generated content is accurate sits with the party presenting it before the court.

For example:

• in the UK case of Anthony Malcolm Cork & Anor v Mark Smith [2026] EWHC 1199 (Ch), a junior solicitor was held to have demonstrated "a serious lack of care and of judgment" when they submitted formal letters to the court in which reference was made to laws that had been hallucinated by an AI program. The judge commented that the junior solicitor "seems to have almost entirely outsourced the thinking process to the program". Even though the employer, a reputable international law firm, had an AI use policy which required all client work, whether produced using AI or not, to be checked in the usual way, the court held that the junior solicitor's managers (i.e. more senior lawyers) remained responsible for the accuracy of the ultimate work product.
• in the Jersey case of Golden Sphinx Limited v Itkin [2026] JRC 138, the Master of the Royal Court issued a note of caution on the use of AI and reminded "all those appearing before the Courts that they are personally responsible for the material placed before the Court". Guidance issued by the Federal Court of Australia (“Litigants in Person Practice Note” dated 5 March 2026) was referred to in this case: "You must check everything you get from AI programs. You must check that all the facts are correct and complete. You must also check that the law is correct." The Master of the Royal Court concluded by saying: "those Australian comments apply with equal, if not greater, vigour in Jersey given the extremely limited amount of material that is available, freely and online about Jersey law."
• in the Guernsey case of The States of Guernsey v Steven Terrence Ogier [2026] GRC 030, it was noted that: "the Court’s guidance that all parties – represented and unrepresented – owe a duty to the court to ensure that cases cited in legal argument are genuine and provide authority for the proposition advanced is of indisputable importance for matters before the Royal Court of Guernsey".

Although the above examples relate to litigation, they are the most public evidence of examples of the impacts of the hazards of ineffective AI use and AI hallucinations. There are likely to be many, many more examples of client reports, research and internal company documents where improper use of AI systems have resulted in erroneous content being relied upon for decision-making.

What employers should do now

Existing policies may not deal adequately with the use or misuse of AI. Employers should review whether they need a standalone AI policy or targeted amendments to existing policies, procedures and employee handbooks.

At a minimum, employers should define which tools may be used, what types of work are suitable for AI assistance, what data must not be uploaded, when human review and sign-off are required, and who is responsible for supervision. Training is also important so that employees understand the limits of AI output and the need to verify facts, authorities and reasoning before relying on it.

Conclusion

Given the pace of AI development, AI use in the workplace is now a current HR issue rather than a future one. Employers should review recruitment processes, privacy information, internal policies and supervisory controls to ensure AI is used lawfully, transparently and with appropriate human oversight.

Used properly, AI can support HR teams. Used carelessly, it can expose organisations to discrimination claims, data protection breaches and poor decision-making based on inaccurate output.

The Employment law team at Bedell Cristin have experience of working in in-house, onshore and offshore roles, and can advise HR professionals on resolving workplace disputes and on updating policies, procedures and employee handbooks on the use of AI.