OpenAI drops NDAs with public officials on its data center projects
Days after midterm voters pushed back on data center expansion, OpenAI has published a community compact. It covers NDAs, electricity costs, water use and hiring reports for the campuses it owns.
OpenAI says it will no longer use non-disclosure agreements with public officials on its data center projects. It announced this on November 5, 2026, as part of a set of commitments to the communities where it builds. The Verge reports that the announcement came in the same week as the midterm elections, in which voters strongly opposed data center expansion.
The NDA change matters because many data center projects have drawn criticism for using such agreements to keep talks between developers and local officials private. Under the new policy, OpenAI says it will allow "only narrow protections covering legitimate security information, trade secrets, or proprietary or competitively sensitive information." OpenAI is not the first to do this. In October, Amazon said it would stop using NDAs with the government agencies it works with on data center projects.
What the compact commits to
OpenAI calls the document a National Community Compact. It applies to future "compute campuses" that OpenAI owns and operates. Compute campus is the company's new name for its data centers. Each campus is to get a "local compact" that includes the national commitments and reflects "local priorities," OpenAI says.
According to OpenAI's announcement, as summarised by FoneArena, the commitments include:
- Electricity: OpenAI will pay "the full cost" of the electricity it uses and of the "infrastructure needed to serve us." Before major spending begins, it will provide financial assurances so that other utility customers are protected if a project is delayed, reduced or cancelled.
- Water: Each campus gets a water plan that prioritises closed-loop or low-water cooling. Water and electricity data will be published for each site.
- Construction and noise: Construction plans will be public. Independent noise studies will be done before operations start and every year after that.
- Jobs: OpenAI will publish how many jobs it expects to create, work with unions and training providers, and report every year on hiring and apprenticeships.
- Engagement: Residents are to be involved early. Each campus will have a named local contact for as long as it operates, and OpenAI will publish an annual progress report.
OpenAI points to two existing projects as examples. In Effingham County, Georgia, it says residents helped shape a $4.8 million grant that pays for free school meals for all 14,000 students. In Pike County, Ohio, families named childcare as their priority, which led to a joint $6 million commitment from OpenAI and SB Energy.
A blueprint aimed at the rest of the industry
OpenAI also published a Compute Campus Policy Blueprint. It is a list of proposals for federal, state and local rules that would apply to all developers, not just OpenAI. The company says voluntary commitments are only a starting point and that policymakers should set clear rules.
On transparency, the blueprint proposes that developers publicly report water, electricity, noise and hiring data. It says states should narrowly limit when officials can be asked to sign NDAs. It also says governments should disclose the terms of tax incentives, attach measurable obligations to them, and be able to take benefits back when developers fall short.
On electricity, it proposes that state utility commissions require large customers to pay for the costs they create and protect other ratepayers if a project fails. It also proposes that utilities make it easier to connect campuses that can safely cut demand when supply is short. OpenAI says its Effingham County campus has an agreement with Georgia Power to reduce demand by up to 1,000 megawatts during periods of high demand. The blueprint also asks Congress to speed up transmission upgrades and to restore and expand tax incentives for clean electricity.
What is and is not covered
The compact applies only to future campuses that OpenAI owns and operates. Much of the infrastructure behind the company's models is built and run by partners. Facilities like those are not explicitly covered by these commitments. All of the pledges are the company's own. In the announcement, OpenAI does not describe any outside body that would enforce them, other than its own public reporting and whatever terms each local compact contains.
The timing is part of the story. Opposition to data centers has moved from public hearings to the ballot box. Developers now face a public that can block or slow projects directly. OpenAI's response is to offer the things that critics have been asking for: disclosure, cost protection for other ratepayers, and measurable local benefits. It is also proposing that those things become rules for every developer.
What it means for work
For local officials, the change removes a constraint. They can discuss OpenAI projects with residents, apart from the narrow carve-outs, and they get regular data on water, power, noise and hiring to check against what was promised. For residents, the published job forecasts and annual hiring reports make it possible to test claims about local employment, which have often been vague in data center deals. For building trades and community colleges, the blueprint proposes using hiring forecasts to train people for specific roles, including electrical work and long-term maintenance of power and cooling systems. Developers would have more reporting and negotiating work at every site if the blueprint's proposals became law. Still open: how far the commitments reach beyond campuses OpenAI owns, who checks the annual reports, and whether other developers follow Amazon and OpenAI without being required to.
Sources
- OpenAI is stepping back from NDAs for data center projectsThe Verge
- OpenAI announces National Community Compact, policy blueprint for AI compute campusesFoneArena
This article was researched, written and illustrated by AI, and published without human editing. Spotted an error? Let us know.
More from māchine
All storiesThe FAA's $875 million SMART system now gives routing and sequencing advice across all US airspace. Airlines back it. The controllers' union says it was never consulted.
Google says its new model scores above its larger Gemini 4 Argon on one index and costs $1.50 per million input tokens. It is meant to carry out the work while bigger models do the planning.
Nvidia's board accepted Huang's resignation on Wednesday. CFO Colette Kress is now interim CEO, and the company has no clear successor as US-China tensions over AI chips grow.