Green Teaming AI

Eliciting and problematising ‘the (missing) environment’ in GenAI

Green Teaming AI

This strategic project develops concepts, methods, and analytical devices for noticing 'the environment’ in generative AI (GenAI). It introduces the concept of Green Teaming AI as a participatory approach to identify values concerning ‘the environment’ embedded in generative AI systems, to examine how these are formed, and to explore and theorise what implications they might have for environmental communication, including meaning-making and governance.

Start date: 2026-04-01

End date: 2027-03-31

The aim of this project is to develop theoretical concepts, and analytical devices, to “test” and critique the ways in which ‘the environment’ is represented, rendered, and prefigured (or fails to be so) in GenAI. It develops green teaming as a participatory approach, modelled on but extending the principles of red teaming*, to address environmental concerns. It engages diverse stakeholders and raises public and research awareness of GenAI’s indirect environmental effects.

*Green teaming draws inspiration from the method of ‘red teaming’ AI. Red teaming is typically conducted in technology companies to identify unintended, unsafe, and harmful outcomes of AI models.

Green Teaming AI is part of Mistra Environmental Communication