By late April 2026, the UK government is still maintaining the position set out in its 18 March Copyright and Artificial Intelligence report: the previously favoured route of a broad text and data mining (TDM) exception combined with a rightsholder opt-out is no longer its preferred option. In practical terms, that means the UK has stepped back from giving AI training a fast, low-friction copyright pathway through legislation.
Just as important, the government has not moved to the opposite extreme. It has openly said that it currently has “no preferred option” and is instead leaning toward more balanced tools such as market-led licensing, training transparency, technical standards and AI output labelling. For the creative industries, that is at least a temporary brake on the idea of “broad exception first, corrections later”; for AI companies, it means the next phase of competition will depend not only on model capability, but also on how well they can evidence data provenance, rights clearance and credible compliance.
