We have spent a lot of this summer talking about what data centers take from our water. A new policy brief from The Water Center at the University of Pennsylvania flips the lens and asks the other half of the question: how can the water sector put artificial intelligence to work, and do it responsibly?
Released in August 2026 and titled “AI for Water: Policy at the Speed of Technology,” the brief distills a full day of conversation from the Water Center’s Spring 2026 Water Policy Forum, which gathered roughly 180 water professionals, utility leaders, academics, software companies, nonprofits, and regulators. It is a clear-eyed, practical read, and it lands right in the middle of the questions our own communities are wrestling with.
Two sides of the same coin
The brief makes a distinction worth holding onto. There is “AI for water,” meaning using AI to help run and protect water systems, and there is “water for AI,” meaning the strain that AI infrastructure like data centers places on our water and energy. The forum focused on the first, but the Water Center deliberately paired it with a companion webinar on the second, “When AI Comes to Town: Water Infrastructure and Community Readiness.” Their conclusion is that you cannot honestly talk about one without the other. That is a framing this network shares.
Where AI can help, and where it gets risky
Opportunities
- Cutting costs through leak detection and reducing non-revenue water
- Operational intelligence: flood prediction, water main break prediction, water quality monitoring, energy optimization
- Capital planning and prioritizing infrastructure needs
- Customer service and connecting people to assistance programs
- Speeding up routine work so staff can focus on judgment calls
Challenges
- Data privacy and cybersecurity
- Workforce hesitation and public distrust
- Uneven governance across large and small utilities
- Lack of training, readiness, and high-quality data
- Funding gaps that leave small, under-resourced systems behind
In his keynote, former New York City environmental protection commissioner Rohit Aggarwala sorted AI uses into four categories, from the most obvious (customer service chatbots) to the most sensitive (automating treatment plants), with a clear message: start where the risk is low and the value is real, and be very careful at the core of operations.
The recommendations that matter most
The heart of the brief is its path forward. A few recommendations stand out for community water advocates like us:
- Keep humans in charge. AI should support decisions, not make them. People should hold final judgment and authority over every AI use.
- Move cautiously and in phases. Start with a readiness stage and a pilot, and prioritize high-value uses like asset management, HR, and leak detection.
- Communicate openly. As public stewards, water providers earn trust by being transparent about how and why they use AI.
- Weigh the equity impacts. Any use of AI has to reckon with the water and energy its infrastructure consumes, and the risk of leaning on AI outputs over community knowledge.
- Build shared standards and treat data as infrastructure. Industry-wide guidelines and well-protected data are what let smaller utilities adopt AI safely.
Why this matters for our watersheds
That equity point is the thread that ties this brief to everything else we have been covering. The same technology that could help a small utility find leaks or predict floods is also driving the data center boom that is reaching for our rivers. The brief’s call to address “AI for water” and “water for AI” together is exactly the balanced, eyes-open approach our communities need. For any group thinking about how technology fits into the future of clean water, it is a strong place to start.
Summarized from “AI for Water: Policy at the Speed of Technology,” The Water Center at the University of Pennsylvania, August 2026. Survey figures reflect responses from forum participants as reported in the brief.
