B-Tech Generative Artificial Intelligence Project Publishes Three Foundational Papers

...The UN Guiding Principles on Business and Human Rights (UNGPs) — the global authoritative standard for preventing and addressing business impacts on people — can add considerable value to efforts aimed at achieving responsible development and deployment of generative artificial intelligence (generative AI) foundation models, applications and products. Based on six months of research and consultation, the UN Human Rights B-Tech Project has identified three broad headlines and associated practical recommendations for how lawmakers, standard setters, businesses and civil society can leverage the UNGPs to foster governance and business practices capable of tackling human rights impacts and risks of generative AI.

Rights-based approaches focus attention on specific harms to people’s dignity and equality. They also provide agreed norms for assessing and addressing impacts, along with a shared language that can facilitate understanding and engagement across diverse stakeholder groups. To catalyse greater attention to applying a human rights lens to developing and deploying generative AI, B-Tech has developed a Taxonomy of Human Rights Risks Connected to Generative AI.

This includes companies that are suppliers of AI knowledge and resources, actors in the AI system lifecycle, and users/operators of an AI system. A UNGPs-informed approach emphasizes that: States should implement a “smart-mix” of regulation, guidance, incentives, and transparency requirements — all supported by policy coherence in domestic and multi-lateral efforts — to advance corporate responsibility and accountability for human rights harms...

B-Tech has published three foundational papers on generative AI:

• "Foundational Paper on "Advancing responsible development and deployment of Generative AI"
• "Taxonomy on Generative AI Human Rights Harms”
• "Overview Company Practice Human Rights and Responsible AI”

Key highlights include:

• Impacts on internationally agreed human rights should be the focus of State and company action to advance the responsible development and deployment of generative AI technologies.
• The UNGPs offer guidance on how to establish the multi-layered architecture of governance needed to address the conduct of private sector actors across the full generative AI value chain.
• Implementation of thorough human rights due diligence by companies developing advanced foundation models will provide an important basis for risk management across the generative AI value chain.

...Recommendations from the first phase of B-Tech Generative AI project are meant to support multi- stakeholder dialogue and collaboration that advances UNGPs-consistent public policy, regulation and business practice. The findings, and responses to them, will inform B-Tech's ongoing work on generative AI in 2024...