The OECD has today published its Due Diligence Guidance for Responsible AI, a piece of work that TUAC supported from the outset but which ultimately fails to deliver on its potential. AI development and deployment presents a wide range of risks for workers, raising new issues for enterprises implementing the MNE Guidelines. Trade unions had expected the guidance to set out concrete actions for addressing these risks, illustrated through clear examples.
Instead, the guidance explicitly adopts a “risk-agnostic approach.” This is disappointing, particularly in light of the growing body of research highlighting human rights impacts at various stages of the AI value chain. Had the guidance included concrete recommendations on specific risks, workers and their unions, along with other stakeholders, could have used it to identify gaps between OECD standards and actual practice and to work with companies to close these gaps. Instead, without discussion of the risks raised by AI or how these should be addressed in line with the MNE Guidelines, the guidance is little more than a collection of broadly available information.
TUAC particularly regrets that the guidance makes no reference to the numerous risks posed to workers by AI in the workplace, and fails to stress the need for AI deployment to comply with the ILO Declaration on Fundamental Principles and Rights at Work. Trade unions also strongly challenge the failure to acknowledge labour rights violations occurring across the AI value chain, and instances of AI being used to undermine freedom of association.
Equally, though meaningful stakeholder engagement is central to the MNE Guidelines, the new guidance’s approach is limited to just information sharing and consultation – with no mention of the need for stakeholders to actively monitor risks and participate in decision-making around mitigation measures. Nor does it acknowledge the importance of collective bargaining and specific clauses in collective agreements in regulating AI.
TUAC hopes that these gaps will be addressed in subsequent OECD work on RBC in relation to AI.
We urge the OECD to properly acknowledge and address the risks AI poses to workers and to actively promote the importance of labour rights at all stages of AI development and deployment. Clear guidance is needed on the steps expected from enterprises to mitigate human rights harms and ensure compliance with existing OECD standards in relation to AI.
Read the full OECD Due Diligence Guidance for Responsible AI.
