Creative Shoplifting
The UK government consultation regarding “updating” copyright law is currently underway. The government seems intent on pushing for changes that water down protections for copyright holders.
The proposals put the onus on copyright holders to opt out of their work being copied and used to train large language models for artificial intelligence. This is like asking Tesco to opt out of shoplifting.
Book publishers have questioned the change, though this has not stopped some publishers from taking advantage of outdated author contracts and licensing electronic versions of authors’ works – without the authors receiving any payment in return.
The creative industries generate £126 billion per year for the UK company – the vast majority of this revenue is accounted for and taxed. In contrast, money made in the future by AI firms will be taxed (if at all) in other countries, such as the US and China.
This is, in effect, a huge transfer of wealth (and effectively) ownership from a taxable industry that employs thousands to a handful of companies that create wealth for a few.
It is simply not possible for copyright holders to know if their work has been used (stolen), and expecting AI companies to be honest and forthcoming is, at best, naïve.
The UK government is rightly concerned with growth, though it appears to be growth at all costs.
AI companies can easily afford to license work for use and training. No industry should be above the law of copyright. AI companies have already hoovered up all the content on the Internet and are now stuck in the position of eating their own dog food – scraping AI-generated content in order to train AI models. This points to a situation where returns are diminishing rapidly. To address this, some AI companies are now training their models using “synthetic” data. This is data created artificially by another AI model.
The irony of this points to the value and uniqueness of creative works – the UK government and others must not legalize appropriation (theft) in order to gain favour with tech companies.