Back

When Gen AI developers made models freely available to the public, knowledge is usually distributed and competition is enhanced. But this openness comes with risks: malicious actors can evade developer oversight, remove any safeguards and customise models for harmful purposes.

This “openness challenge” is complicated by open washing: companies misleadingly brand models as “open-source” or “open", while nonetheless maintaining use and access restrictions. This undermines the public’s understanding and create diversions from Gen AI risks.

The grantees advocate for policy that fosters openness towards external researchers, to capture the full benefits of transparency. Engaging diverse experts during development and post-release allows the AI community to better manage the interplay of risks and benefits of Gen AI.

Barriers exist that limit the potential of openness to external researchers. For example, developers can limit the types of information and the model elements that they make available to researchers, limiting the scope and robustness of research.

Current policy approaches do not adequately tackle the openness challenge: Neither EU, UK nor US frameworks adequately tackle the openness challenge of Gen AI models. Yet, the EU has an opportunity to tackle this issue through the AI Act Code of Practice on General Purpose AI.

he paper therefore recommends a set of non-mutually exclusive policy options that could be taken on board by countries who are drafting and/or reviewing their own AI legislation.

Access to results and contact: Cutting through open-washing - https://infrastructureinsights.fund/projects/cutting-through-open-washing/ // https://infrastructureinsights.fund/wp-content/uploads/2023/11/AWO-2024-Generative-AIs-open-source-challenge-policy-options-to-balance-the-risks-and-benefits-of-openness-in-AI-regulation-041124.pdf

- Huge Congrats on the release of your report!