As the healthcare sector rapidly adopts and deploys artificial intelligence, one organization urges stakeholders to use their purchasing power to demand environmental transparency from AI vendors.
Health Care Without Harm (HCWH) has published its U.S. Artificial Intelligence Procurement Guide to support healthcare organizations in identifying their preferred AI vendors and encouraging them to become better stewards of the environment. The guide was created at the behest of HCWH’s health system and payer partners, who were also involved in its development.
There are many environmental impacts of AI, including air pollution, energy disruptions, rising electricity prices, community displacement, reduced water availability, noise pollution, waste and lack of green space access, HCWH’s guide notes. Healthcare stakeholders have an opportunity to wield their collective influence to demand protection of the planet and human health.
“We want to be able to leverage the buying power of the healthcare sector to make all Americans, all people in this country, healthier,” Jenny Keroack, director of climate solutions for HCWH’s U.S. Climate Program, told Fierce Healthcare. “We want to see AI companies become net assets for their communities.”
The guide recommends healthcare organizations ask AI vendors to measure and reduce their greenhouse gas emissions, disclose their energy mix and increase renewable energy use and reduce the environmental footprint of AI models.
It also recommends AI vendors be transparent about data center energy efficiency, water use and backup generators; account for community and public health impacts; and help healthcare orgs reduce the footprint of their own AI use.
“Procurement is not only a tool for gathering this data. It’s also a tool for eventually putting obligations into contracts,” Emily Bancroft, HCWH’s CEO, told Fierce Healthcare.
The guide features 10 environmental disclosure questions about an AI vendor’s company, models and data centers. It includes boilerplate language for healthcare organizations to use when purchasing, as well as scoring guidance for the responses. Vendors can score 25% the first year for committing to provide data within 12 months the first time they respond to the questions. This allows leeway for vendors who may need time to collect data on third-party AI infrastructure, like data centers. After the first year, commitments don’t receive points and vendors must provide the requested data.
“What we’re asking for here isn’t something that’s impossible,” Bancroft said. “You can build your data centers in ways that are more environmentally friendly, that reduce pollution, that protect the water sources near them.” But this, she added, requires collective demand and healthcare leaders recognizing that their investment gives them influence.
The guide cites the Sustainably Advancing Health AI (SAHAI) Framework, published in The National England Journal of Medicine in 2025 by an interdisciplinary team across Stanford and Cornell Tech. SAHAI proposes solutions like scheduling nonurgent processing during off-peak hours when grid energy demand is lower and leveraging mixed-quality models that rely on simpler models for initial data processing and more complex ones for high-performance tasks like drafting responses to patients.
For products that have optional AI features, organizations can request to turn those off in order to avoid racking up unnecessary emissions and costs. Energy efficiency should save money for all parties involved, Keroack notes. “The things we’re talking about can reduce cost as well as being more environmentally friendly,” Keroack said.
The guide grapples with an industry imbalance when it comes to Science Based Targets initiative (SBTi) commitments. SBTi accelerates science-based corporate climate action, aligning around goals like hitting net-zero by 2050. Over 940 companies in the U.S. have validated science-based targets, many of them large healthcare orgs, Keroack said. Yet the same is not yet true of AI companies, especially smaller ones. The guide is meant to support that.
Bancroft pointed to successful past efforts in healthcare that required collective action and supply chain involvement, such as the phasing out of harmful chemicals in furnishings or mercury in medical devices. States, too, have a role to play in regulation, Bancroft noted, particularly as the Trump administration reverses climate regulations—most recently, a historic repeal of limits on fuel-fired power plant emissions.
“Healthcare has the opportunity, but also an obligation because of its mission, [to say] health isn’t optional,” Bancroft said, “and AI vendors must guard the environment and wellbeing if they want to get our business.”