From the journal

UK Crime and Policing Act 2026 Criminalises Unsafe AI Chatbot Supply, United Kingdom, April 2026

The UK Crime and Policing Act 2026 received Royal Assent on 29 April 2026. Section 255 creates three criminal offences for AI chatbot suppliers: deploying a chatbot that produces illegal content, failing to conduct an adequate risk assessment, and failing to mitigate identified risks of harm. Convictions on indictment carry a maximum sentence of five years' imprisonment. The Act brings AI chatbot providers within scope of the Online Safety Act 2023 illegal content obligations.

2 min read

The UK Crime and Policing Act 2026 received Royal Assent on 29 April 2026. Section 255 of the Act creates criminal offences for suppliers of AI chatbots that produce illegal or harmful content. The provisions are enacted law in force from the date of Royal Assent. AI chatbot providers are now within scope of the illegal content rules in the Online Safety Act 2023.

Section 255 of the Crime and Policing Act 2026 defines three distinct offences. The first is supplying an AI chatbot that produces content specified in subsection (4). The second is failing to carry out a suitable and sufficient risk assessment before supplying such a chatbot. The third is failing to mitigate or manage risks of identified harm. Specified content means content that is illegal for any user under section 59 of the Online Safety Act 2023. For users under 18, the category extends to content where an AI falsely presents itself as human. Penalties reach five years' imprisonment on conviction on indictment.

AI chatbot developers, API providers, and platform operators deploying conversational AI to UK users must assess whether their systems can generate content caught by the specified content definition in section 255(4). Operators serving users under 18 face broader obligations: their systems must not produce content designed to mislead children into believing they interact with a human. Third-party AI model customers are classified as suppliers under section 255 and bear direct liability rather than downstream platform liability. Providers must document their risk assessments to demonstrate compliance in any enforcement proceeding.

Section 255 does not include a statutory safe harbour for suppliers who conduct risk assessments that prove insufficient after the fact. Ofcom guidance on what constitutes a suitable and sufficient risk assessment for AI chatbots under the combined Crime and Policing Act and Online Safety Act regime is pending. The boundary between platform liability for user-generated content under the Online Safety Act and direct supply liability under section 255 has not yet been addressed by regulators.

Licentium advises technology companies and AI product developers on UK regulatory obligations. If your AI chatbot products serve UK users, we can assist with compliance analysis and risk assessment documentation. Work we undertake includes Online Safety Act compliance, AI product risk assessments, criminal liability analysis for technology products, and regulatory engagement strategy.

Source: Crime and Policing Act 2026, s 255 (UK c.20), Royal Assent 29 April 2026

AI Regulatory

More from the journal

See all
Illia Prokopiev

From Cloud Concentration to AI Dependence: The UK’s Critical Third Parties Regime

The United Kingdom now directly oversees designated technology suppliers whose service failures could threaten financial stability. The question is whether the first cloud designations show a legal expansion toward AI-model providers, and what the present regime requires. This analysis assumes the quoted statement concerns the UK financial-services Critical Third Parties regime and assesses the law through 14 July 2026.

Alberta Regulated iGaming Market Launched on 13 July 2026 with 22 Operators

Alberta's regulated private iGaming market launched on 13 July 2026, making Alberta the second Canadian province to permit private online gambling operators after Ontario. The Alberta Gaming, Liquor and Cannabis Commission serves as market regulator and the Alberta iGaming Corporation oversees commercial operations and operator contracts. Twenty-two operator sites went live on day one, including FanDuel, DraftKings, BetMGM, and BetRivers. Operators must fully launch or exit the Alberta market by 13 October 2026.

European Commission Presents Cybersecurity and AI Action Plan on 7 July 2026

On 7 July 2026, the European Commission presented an Action Plan on Cybersecurity and Artificial Intelligence. The plan directs the Commission and ENISA to evaluate advanced AI models before they reach the EU market, establish a secure testing platform for critical-sector organisations, and launch an EU Grand Challenge on AI-powered cybersecurity solutions. It operates alongside the AI Act, NIS2 Directive, DORA, Cyber Resilience Act, and Cyber Solidarity Act, and introduces no new directly binding obligations.