UK Government responds to House of Lords report on Gen AI
May 9, 2024
On 17 April, the UK Government responded to the February 2024 report from the House of Lords Communications and Digital Committee on large language models and generative AI.
There is information about that report here. It included comments and recommendations regarding copyright compliance and transparency.
You can see the UK Government’s response here.
It includes the following on copyright and transparency:
The Government is committed to supporting, not undermining human creativity, and developing an approach to AI and copyright that allows the AI and creative sectors to grow together in partnership.
…
Examples of market-led approaches to data licensing are emerging across the AI and creative sectors, with deals being struck to enable access to high quality data on mutually agreed terms. Collective licensing is one approach that can provide at-scale access to data and can enable creators to be remunerated for their work. The Government is also working with creative and journalism sectors to better understand opportunities for greater access to expanded, high quality data sources at the scales needed for LLM training.
Transparency over the use of copyrighted material to train models and regarding attribution of outputs is important. The Government believes that greater transparency is needed from AI developers in relation to the data used to train LLMs so that rights holders can better understand whether content they produce is used as an input into AI models. We are separately engaging with stakeholders to understand broader perspectives in relation to transparency about the purposes of web crawlers. Transparency over training data has a particular significance in an IP context, but also in a broader safety context, for example, transparency about bias in training datasets. Greater transparency is also needed around the attribution of outputs. In its response to the AI regulation white paper consultation, the Government confirmed that it intends to progress work on transparency – working with both rights holders and AI companies to understand what is technically feasible and what is proportionate.
In Australia, these issues are being considered by the Copyright and Artificial Intelligence Reference Group (CAIRG), which includes Copyright Agency, Australian Society of Authors and Australian Publishers Association: see further here.