Skip to content
As the rise of AI and the technology surrounding it surges, it is vital that we, as people and as owners of Intellectual Property, have protection and a government that we can rely on to help and assist us. In early 2025, the UK government proposed a significant revision to copyright and data usage regulations. The proposal was to introduce an “opt-out” model.
What was the Opt-Out Proposal?
The “opt-out” model proposed by the government would allow artificial intelligence (AI) developers to train and program their AI models on copyrighted works automatically, unless they explicitly opt out of this process. The only way they would not be able to do so would be if the creator and owner of the copyrighted works explicitly objects and opts out of allowing it. This would have given AI developers complete access to any publicly available online content to train their systems. The goal was to present it as a proposal balancing innovation with existing rights protections; however, this approach favoured AI companies and almost bullied individual copyright holders.
The Public Backlash and Withdrawal
This proposal by the UK government was highly controversial and received widespread condemnation. Artists, writers, and musicians felt betrayed and vulnerable; the proposal favoured the AI developers without regard for the copyright holders and individuals. However, the public’s vocal opposition played a crucial role in the government’s decision to abandon the proposal. Thankfully, only a few months later, in May 2025, the UK government decided to leave the proposal due to the ever-increasing backlash and outrage. The government decided that it needs a licence-based system, and a more comprehensive AI regulatory framework proposal will be presented later.
The withdrawal itself is a win for creatives and copyright owners; however, this may not be the end of the issue, as the underlying problem remains unresolved. The very idea of an ‘opt-out’ system remains a real threat, which could re-emerge in the future in many forms. It is essential to understand that while this approach would enable AI companies to thrive and drive the growth of AI industries, it would also have a profoundly negative impact on individuals and copyright holders. The potential impact on the AI industry, in terms of innovation and development, should also be taken into account in any future policy decisions.
Why It Was Never a Good Idea and Why It Should Not Return
Now that the proposal has been shelved, we must understand why it was prematurely and poorly implemented with inherently flawed ideas and ideals.
One central argument against the opt-out model is that it undermines the core principles of the copyright system. Creators of works automatically hold the copyright to their works, and this principle exists to ensure that they can control how their intellectual property is used across the internet. The opt-out model proposed has taken this ideal and flipped it on its head, as instead of making AI companies and models seek out permission to use copyrighted works, the owner of the copyrighted work must proactively retract their assumed consent. This assumes that the copyright owner is aware that their copyrighted work is available for these AI companies to use and is vulnerable; otherwise, they would not need to opt out manually. This idea and proposal, based on this principle, is an attempt to redefine and reduce the ownership and control an individual has over their own copyrighted work, which blatantly opposes the concept of copyright and intellectual property.
As stated in the previous paragraph, how will a copyright owner know that they must opt out of their copyrighted work being used, or if it is even being used in the first place? This places an impossible burden on creators, as it imposes unrealistic and unfair expectations on them not only to monitor and manage their work, but also to police its online use. How would a freelance illustrator know that their drawings were scraped into an AI company’s dataset and being used to train an image generator like Midjourney? The question of where to submit this opt-out request arises, and would opting out apply retroactively to models that have already been trained using your copyrighted work?
This leads to my next point: there is little to no transparency about what these AI companies are using. One of the most critical points of criticism levied against AI companies is that the government does not force them to disclose which copyrighted works they have used to train their AI models. There have been multiple attempts to introduce mandated transparency, but these have always been blocked by the ministers in the Commons, citing the grounds of “financial privilege”. This blatant lack of transparency allows AI companies to evade informing creators that their work is being used, resulting in creators not knowing they must actively object to their work being used.
This proposal risked severe, long-term economic and cultural harm. Allowing AI companies and developers to scrape and learn from copyrighted works without any form of payment, licensing deals, or compensation effectively transfers value from creators to tech companies. This, on a moral standpoint alone, is more than enough reason for this proposal to backfire, but it is also economically damaging. The UK is a country with one of the largest creative industries in the world, and its creative industries contribute over £100 billion to the economy annually. However, the proposal undermined the financial foundation of these industries, enabling free AI training, and risked severe, long-term economic and cultural harm.
These points have collectively eroded the trust of AI companies and the government, as despite repeated warnings from two cross-party parliamentary committees arguing that the opt-out model was unworkable and unfair, the government remained optimistic that the proposal would be effective. This was, of course, until it was announced, and the public backlash prompted them to backpedal. Many people already view generative AI tools with scepticism, especially those that create images or videos via a prompt that closely mimics or remakes human-created content. The idea that such AI companies were training their models for financial benefit without the consent, notice, and without compensating the material the AIs used to learn, struck a nerve with many people. Had the opt-out system gone ahead, it would have further eroded the trust that the public has not just in AI, but also in the government’s ability to regulate it reasonably.
Final Thoughts
Now that the government has shelved the opt-out proposal, it has stated that it will develop a far more comprehensive AI bill in the future, which is said to be built on sustainable, inclusive, and ethical foundations, without the downsides of the previous proposal. This underscores the need for a more balanced approach to AI regulation, one that considers the rights of creators and the potential of AI technology. We must not forget that this was not just a mistake or misstep, but a perilous precedent, which, whilst it has been abandoned, could easily resurface at any point or in different forms. It is up to all of us, as creators, consumers, and policymakers, to ensure that future AI regulations are fair and just.
