A more focused way of working

• The group agreed to organise monthly meetings around a single deep‑dive topic and to develop a shared knowledge repository to capture and reuse insights across the sector.

Progress on supplier engagement

• Early trials of the supplier engagement document are underway, particularly within technology procurement, with a commitment to revisit experiences and outcomes later in the year.

AI emissions under the spotlight

• Participants compared methods for measuring AI‑related energy use, including analysis of cloud usage data and token‑based functional units, highlighting the difficulty of separating AI from non‑AI consumption.

Greater nuance in AI energy intensity

• Discussion explored how different phases of model operation, as well as prompt and output length, can significantly affect energy use, underscoring the limits of single averaged metrics.

Toward practical benchmarks

• There was strong interest in combining real‑world cloud data with token‑based approaches to establish realistic emissions ranges for AI systems operating at scale.

Building the evidence base

• Progress on AI case studies for an industry white paper was reviewed, with a need identified for more publisher‑specific examples to complement existing research.

Improving transparency from AI providers

• The group agreed to collaborate on questions and recommendations to encourage clearer disclosure of energy and emissions data from commercial AI model providers.

Clear priorities confirmed

• Five focus areas were reaffirmed for the year ahead - user behaviour, supply chain engagement, artificial intelligence, regulation and compliance, and print versus digital strategy - with AI remaining a standing agenda item.

The session laid the groundwork for a more structured and collaborative programme of work, beginning with a February deep dive on regulation and compliance.