Sanctions Frameworks for AI-Related Misconduct: Who Is Really to Blame?

AI is a double-edged sword. Despite its many benefits, it comes with serious risks if the information is not accurately checked and verified. This can lead to sanctions from the relevant regulator, the courts, and in some cases even the police.
 

AI is now commonly used by professionals in the workplace, as it makes tasks more efficient, and minimises errors. However, AI can fabricate information or misrepresent facts which are presented by the technology as genuine. This phenomenon is known as AI ‘hallucinations.’ The work AI produces therefore needs to be carefully checked by professionals to ensure that it is in fact accurate. It is the responsibility of the professional using the AI to check the information the AI tool has produced. The failure to check and verify the information can lead to referral to the relevant regulator, reputational damage, disciplinary action, and potentially even claims of negligence.

The potential sanctions for AI related misconduct for a number of different professions are discussed in more detail below.  

Law
 

Over the last 12 – 18 months, reports have begun to appear in the media for the misuse of AI by professionals. At the time of publication, much of these matters relate to legal professionals, but the principles from these cases apply for professional individuals across several sectors and professions.

The case of Ayinde v Haringey [2025] EWHC 1383 (Admin) concerned judicial review proceedings. A pupil barrister working on the case submitted submissions contains fake citations and misstatements of law. The pupil barrister denied using AI, however the Divisional Court said that the pupil barrister “had not provided to the court a coherent explanation for what happened.” It was later discovered that the pupil barrister had previously made submissions which had also included false citations, and the matter was not referred to the regulator, the Bar Standards Board, due to assurances given by the pupil barrister, and the Head of Chambers. Embarrassingly, the issue arose again in the Ayinde case, showing the reputational damage to the barrister, and indeed the Chambers involved, for failing to check information produced by AI.

The case of Al-Haroun v Qatar National Bank [2025] EWHC 1383 (Admin) involved a damages claim for breaching a financial agreement. The solicitor in question relied on the legal research of his lay client (who had utilised AI) without himself verifying it as accurate. The Claimant sought to rely on a number of authorities that were fictious, or did not support the arguments put forward. The solicitor admitted that he had sought to rely on a number of fictious authorities, and apologised to the court for this error. The solicitor was later referred to the regulator, the SRA.

In May 2026, international law firm, Pinsent Masons, was criticised by a Judge during an insolvency case where AI hallucinated statutory authorities, and these were relied on in an initial letter to the Court, without these references being checked. When the Court queried the reference, a second AI-drafted letter was sent. This case has attracted considerable press attention, with the Judge stating that the matter served as a “public admonishment” of the firm and individual solicitors involved. Contempt of Court proceedings were contemplated by the Judge but not pursued. However, the case demonstrates the reputational damage that can be caused by failing to check information produced by AI.

Striking Out of Documents

Courts may strike out pleadings or submissions containing false citations created using AI. This would be embarrassing for any practising lawyer, as they would have to explain to their client why the work produced by AI was not checked thoroughly before being submitted to the court.

Contempt of Court Proceedings

Lawyers may be found in contempt of court for failing to check the accuracy of the information produced by AI. In the Ayinde case set out above, the court ruled that the pupil barrister’s conduct had in fact reached the required threshold for contempt of court proceedings to be brought against them. In the most serious cases, legal professionals may also be referred to the police for perverting the course of justice.

Healthcare
 

In the medical sphere, the NHS is now advising doctors to use AI to generate transcriptions and clinical summaries. AI is now used in hospitals, GP surgeries, and outpatient clinics. The rationale for this is that this will enable doctors and other healthcare professionals to spend more time with patients, rather than on the administrative task of writing up notes. However, there are risks with this approach. For example, if there are transcription and summarisation errors (hallucinations) which are not spotted, this could feature on patient’s medical records, and in turn affect their diagnosis or treatment moving forwards. AI tools may also misunderstand or misrepresent complex medical terms, or they may have difficulty understanding different accents. In healthcare, it is therefore vital that there is human oversight to check the information produced by AI to ensure medical notes are in fact correct. If not, this may lead to adverse outcomes for patients, local investigations, or even referral to the relevant regulator.

Teaching
 

AI can support teachers in planning lessons, summarising resources, and completing various administrative tasks. However, teachers need to be mindful of the quality and accuracy of the information produced by AI. Indeed, the National Education Union has advised teachers that “if you create learning resources using GenAI tools, that you always carefully check, and consider the output, as GenAI will produce factually inaccurate content and will tend to recreate biases in its results.” It is therefore important that teachers check the information produced by AI, as any misuse of AI could affect learning outcomes, or be referred to the regulator, the Teaching Regulation Agency.  

Conclusion
 

The sanctions landscape in this area is quickly evolving. Professionals using AI (in whichever profession) are ultimately the ones who are responsible for the work that the AI tool produces. They therefore need to verify that all the information generated by AI is accurate and correct. If they do not do so, then the sanctions for failing to do this can be severe; and involve reputational damage and referrals to the relevant professional regulator. In more serious cases, the police, and even the criminal courts may be involved.

About the author

Kate is an Associate in the Regulatory Team at Kingsley Napley. She first joined the firm in 2020, and then subsequently returned in March 2024, after completing her Training Contract at a regional firm. Kate was admitted to practise in England & Wales in March 2024.