This week, we got an eagerly anticipated look at how the US Government is thinking about openness and AI, a question that we’re focusing on a lot at Mozilla. On Tuesday, the National Telecommunications and Information Administration (NTIA) published their review of “open-weight” foundation models, and the risks, benefits, and potential policy approaches. NTIA’s report is the result of a public consultation process coming out of the Biden Administration’s Executive Order on AI. Mozilla weighed in with our views, sharing our history in the open source movement and the value of openness in AI.
We welcome NTIA’s recommendation that the US government take no current action to restrict open foundation models; this approach supports the ability of open source AI to flourish, promoting the benefits that openness can bring, like innovation, competition, and accountability.
In their report, NTIA rightly notes that “current evidence is not sufficient to definitively determine either that restrictions on such open-weight models are warranted or that restrictions will never be appropriate in the future.” Instead of recommending restrictions, NTIA suggests that the government “actively monitor…risks that could arise from dual-use foundation models with widely available model weights and take steps to ensure that the government is prepared to act if heightened risks emerge.”
NTIA’s recommendations for collecting and evaluating relevant evidence include support for external research, increasing transparency, and bolstering federal government expert capabilities. We welcome this approach, and in our comments we called for governments to play a role in “promoting and funding research”; we agree that it will help us all better understand and navigate the AI landscape.
Competition is also featured in NTIA’s report. While it’s recognized that open-weight models alone aren’t sufficient to bring about lasting and meaningful competition in the ecosystem, NTIA notes that foundation models can lower market barriers to entry and “decentralize AI market control from a few large AI developers.” Mozilla’s story began when we sought to bring much-needed competition to the browser market. We’ve watched with concern as similar patterns of concentration emerged in the AI ecosystem, and agree that openness in AI can contribute to a solution.
As we noted in our submission to NTIA, “[g]ood policymaking on AI, and on openness in AI in particular, therefore requires a careful balancing of benefits and risks as well as analytical rigor in taking into account the various dimensions and actors in the AI ecosystem. Rash decisions and ill-considered solutions may cause irreparable damage to the ‘open source’ AI ecosystem.”
We appreciate that NTIA has heard these concerns in their work and wholeheartedly support the broader lens through which they considered these tough questions around openness and AI – including recognizing the spectrum of openness, and centering analysis on marginal risk. Building on NTIA’s own recommendations, we also reiterate the importance of providing robust support to the open source AI ecosystem and involving a broad range of stakeholders in making decisions about its governance, including through measures such as:
- Supporting the open source AI community in developing norms and practices around responsibly developing and openly releasing AI models and components.
- Investing in and providing resources for the development and maintenance of ‘open source’ AI.
- Involving federal agencies responsible for protecting civil rights, promoting competition, and advancing scientific research in the development of policy touching on openness in AI.
Mozilla was grateful for the opportunity to share our views as part of this important process. We believe that supporting openness in AI and tackling the challenging questions that the technology presents – including as a partner to policymakers in efforts like this one – can create a safer, more trustworthy, and better AI future for all.