Recent decisions by leading AI labs to either open-source their models or to restrict access to their models has sparked debate about whether, and how, increasingly capable AI models should be shared. Open-sourcing in AI typically refers to making model architecture and weights freely and publicly accessible for anyone to modify, study, build on, and use. This offers advantages such as enabling external oversight, accelerating progress, and decentralizing control over AI development and use. However, it also presents a growing potential for misuse and unintended consequences. This paper offers an examination of the risks and benefits of open-sourcing highly capable foundation models. While open-sourcing has historically provided substantial net benefits for most software and AI development processes, we argue that for some highly capable foundation models likely to be developed in the near future, open-sourcing may pose sufficiently extreme risks to outweigh the benefits. In such a case, highly capable foundation models should not be open-sourced, at least not initially. Alternative strategies, including non-open-source model sharing options, are explored. The paper concludes with recommendations for developers, standard-setting bodies, and governments for establishing safe and responsible model sharing practices and preserving open-source benefits where safe.
Seger, Elizabeth and Dreksler, Noemi and Moulange, Richard and Dardaman, Emily and Schuett, Jonas and Wei, K. and Winter, Christoph and Arnold, Mackenzie and Ó hÉigeartaigh, Seán and Korinek, Anton and Anderljung, Markus and Bucknall, Ben and Chan, Alan and Stafford, Eoghan and Koessler, Leonie and Ovadya, Aviv and Garfinkel, Ben and Bluemke, Emma and Aird, Michael and Levermore, Patrick and Hazell, Julian and Gupta, Abhishek, Open-Sourcing Highly Capable Foundation Models: An evaluation of risks, benefits, and alternative methods for pursuing open-source objectives (October 9, 2023). LawAI Working Paper No. 2-2023, Available at SSRN: https://ssrn.com/abstract=4596436 or http://dx.doi.org/10.2139/ssrn.4596436