Written evidence submitted by Mozilla (DCG0021)

Digital Centre of Government Inquiry

About Mozilla

Mozilla’s mission is to ensure the internet is a global public resource, open and accessible to all. An internet that truly puts people first, where individuals can shape their own experience and are empowered, safe, and independent.

Founded as a community open source project in 1998, Mozilla consists of several organizations. Most notable is the non-profit Mozilla Foundation and its wholly owned subsidiary, the Mozilla Corporation, which leads our market-based work, including the development of the open-source Firefox web browser and Gecko, one of only three major browser engines. They work in close concert with each other and a global community of tens of thousands of volunteers under the single banner: Mozilla.

For the past five years, Mozilla has been committed to advancing trustworthy AI. Mozilla published a paper in early 2024, Accelerating Progress Toward Trustworthy AI, that outlines how Mozilla and its allies are advancing openness, competition, and accountability in AI. Mozilla is putting its resources behind these priorities as well: The Mozilla Foundation has been dedicating 100% of its $30M a year budget to philanthropic activities, advocacy, and programmatic work on this topic. Mozilla is also investing another $30M in research and development on trustworthy AI via Mozilla.ai, as well as $35M in responsible tech startups — including investments in notable startups like Hugging Face — through Mozilla Ventures and the Mozilla Builders accelerator program. On the frontlines of modern AI practices, Mozilla freely provides an open-source, large-language model (LLM) AI model deployment system for local use and empowers more people to enhance the safety of models online through the 0Din bug bounty program. In addition, Mozilla supports organizations engaged in artificial intelligence research, including the Open Source Initiative’s work on defining Open Source AI, and through the efforts of the Columbia Convening which brought together a diverse set of stakeholders to talk about Openness and AI.

As an independent and mission-driven organization, Mozilla is committed to working with regulators to develop effective policies that ensure that innovation and growth in AI serve the public interest.

+++

One of the key responsibilities and opportunities DSIT holds as the ‘digital centre of government’ is the development of the National Data Library. At Mozilla, we are excited to see plans for the National Data Library develop and make the most of the opportunity to unlock the value of public data for AI innovation and deployment in a way that benefits people, improves public services and supports innovation and growth.

In order to achieve this, it is important to ensure openness runs through the initiative. This means creating an open data set that is able to be safely accessed and scrutinised, is produced with engagement from the public and communities, and in a way that evens the playing field for organisations and developers of all sizes.

Open data sets create more opportunities for startups to innovate and scale faster. The UK has masses of public data it could use to power responsible AI growth. For open source AI innovation, particularly, including innovation for the benefit of public services, open data sets can democratise access to AI by providing more people with access to quality data needed to reduce harmful and biased content and fine-tune models. Opening data sets will also enable researchers and civil society to scrutinise models and help ensure better quality data.

Talking to startups and other developers, it is clear that too often UK public data today is locked away, silo-ed or otherwise corrupted. To make a National Data Library AI-ready, we recommend the Government focus on pulling together data from departments and agencies and institute open, interoperable standards to safely share data and set up clear governance structures that can enable innovation whilst engendering public trust.

We would also recommend the Government offer departments and agencies targeted support to improve their data sets – enabling them to more easily and regularly clean, standardise and share data.

In line with this, any AI related legislation should support Public AI infrastructure, whether by aiming to make data more open, providing more accessibility to AI solutions, supporting the open-source movement, or giving individuals the tools and education needed to leverage AI for their own benefit.

Any legislation should acknowledge that, like the internet, AI should become a global public resource as it is built from data garnered from humanity’s collective intelligence, and that it should be open and accessible whenever possible.

Our believe in the power and importance of open data was part of the reason behind our recent Columbia Convenings - bringing together over 40 AI experts to discuss the meaning of openness in AI and discussing how to enable trustworthy and beneficial AI. Part of this examined technical data standards for openness which we would be happy to share.

As plans for the National Data Library and DSIT’s role as the digital centre of government continue to develop, we would be pleased to work with this Committee and the Government to ensure the initiative has the biggest impact it can, particularly on the issues of openness, open-source AI innovation, and open-standards.

03 March 2025