Speed is the point. Standard UK primary legislation takes years, Green Paper, consultation, White Paper, Parliamentary scrutiny, Royal Assent, commencement. The Children’s Wellbeing and Schools Act (CWSA) and the Crime and Policing Act (CPA) received Royal Assent on April 29, 2026, and in doing so created a different kind of regulatory pathway for AI chatbot oversight: a Secretary of State power to move faster than primary legislation allows.
According to legal analysis from Bird & Bird, the legislation introduces new powers – reportedly through a Section 214A addition to the Online Safety Act, allowing the Secretary of State to restrict children’s access to “specified functionalities” of AI services. The section number itself is cited from legal analysis, not confirmed against the official legislation.gov.uk text. Readers relying on specific statutory citations for compliance purposes should verify against the primary source.
That qualification matters for lawyers. The core development is confirmed: new powers exist under UK law as of April 29, 2026 to restrict what AI chatbots can do for users under 18. The mechanism is executive action through secondary legislation under the OSA framework, faster and more targeted than a new primary Act, and adjustable as AI capabilities evolve.
The scope of “specified functionalities” will be defined by secondary regulations. That’s deliberate. A primary Act that tried to enumerate every AI chatbot feature worth restricting would be obsolete before it passed. The CWSA approach designates the power; subsequent regulations define its exercise. For AI chatbot operators, that means the immediate compliance question isn’t what’s restricted now, nothing specific has been restricted yet through this mechanism, but what monitoring and response processes are in place for when restrictions come.
The Act reportedly requires the government to publish a progress statement on the first regulations by approximately July 29, 2026, according to analysis of the legislation. That’s roughly 90 days from Royal Assent, a standard legislative review timeline. July 29 is the earliest point at which the market gets meaningful signal on what “specified functionalities” will mean in practice.
For context: the Washington State AI companion chatbot law and similar US state measures have created jurisdiction-specific requirements for AI services accessible to minors. The UK CWSA approach differs in that it creates a flexible executive power rather than specific statutory prohibitions, it’s a governance architecture, not a rule list. That flexibility cuts both ways: it allows the government to respond quickly to harmful AI applications, and it means operators face regulatory uncertainty until secondary regulations are published.
AI chatbot operators serving UK users under 18 should be tracking this legislation. The immediate action isn’t compliance with a specific restriction, it’s ensuring the product and legal teams are monitoring the July 29 progress statement and any subsequent consultations, and that the organization has a process for responding to secondary regulations within their likely implementation timeline.
*Human verification recommended: The specific statutory section number (Section 214A) cited in legal analysis is pending confirmation against the official legislation.gov.uk text. Readers relying on specific statutory citations should verify against the primary source before publication or legal use.*