The UK government withdraws a proposal to let AI companies train on copyrighted works unless creators opt out, after backlash from artists like Dua Lipa
The UK government has backtracked on its position on copyright and AI, stating it must take time to “get this right”.
Context & Ripple Effects
The withdrawal follows a repeated UK struggle to settle rules for AI training on copyrighted material. The government had already [[a:1164714|delayed planned copyright changes after its consultation failed to produce support for any proposal]], echoing an earlier episode in which it shelved proposed AI-training rules after artists and technology groups could not agree.
The abandoned opt-out approach had been put forward through the UK’s copyright consultation, which also contemplated greater model scrutiny. Opposition was not confined to creators: OpenAI and Google rejected the proposed opt-out framework, underscoring how difficult it is to build a rule that both rights holders and AI companies accept.
First-order effects
- Creators retain the existing position rather than having to actively reserve rights under the proposed opt-out regime; the government must return to the policy design process.
- AI companies lose the prospect of a near-term UK exception structured around creator opt-outs, leaving the status of a new training-specific rule unresolved.
Second-order effects
- The reversal strengthens creators’ leverage in the next consultation or legislative process, since public opposition helped stop a proposal rather than merely amend it.
- Model developers and rights holders have less regulatory clarity for UK-facing training and licensing decisions while the government revisits the framework.
Third-order effects
- The episode suggests AI-copyright policy will be shaped as much by the practical allocation of consent and bargaining power as by a broad desire to support AI development.
- If repeated withdrawals continue, the UK may move toward narrower, negotiated rules rather than a single default opt-out exception—though the eventual approach remains open.
The trend: This is one data point in the shift from broad AI-training exceptions toward contested frameworks for creator consent, transparency, and content licensing.