OpenAI Data Center NDA Ban for Public Officials

OpenAI data center NDA practice is changing: the company says it will not use nondisclosure agreements with public officials on data-center projects, and it has published a wider set of community commitments after midterm voters pushed back on large computing sites. The Verge's Jay Peters reported the shift on 6 November 2026, noting that only narrow protections for security and trade secrets would remain.

The policy rewrite comes as communities weigh jobs and tax base gains against power demand, land use and water concerns. OpenAI's answer, as summarised by The Verge, mixes transparency promises with a pledge to shoulder infrastructure costs that critics say too often fall on ratepayers.

What the OpenAI data center NDA change covers

Under the new stance, public officials would not be asked to sign broad NDAs merely to discuss or negotiate data-center plans. Limited confidentiality could still apply where genuine security needs or trade secrets are at stake. That carve-out is meant to protect sensitive technical details without muffling ordinary civic debate about siting, subsidies and environmental impact.

OpenAI also says it will pay the full cost of electricity and infrastructure required for its projects, publish expected job counts and hiring reports, and — for future campuses it owns — adopt a local compact that reflects community priorities. Those commitments turn abstract "good neighbour" language into a checklist local governments can cite in hearings.

• No broad NDAs with public officials on data-center projects
• Narrow exceptions only for security and trade secrets
• Company to cover full electricity and related infrastructure costs
• Public reporting on expected jobs and hiring
• Owned "compute campuses" to use a local compact tied to local priorities
• Context: midterm voter opposition to data centers; Amazon said in October it would stop NDAs with government agencies on data centers

Why midterm politics forced a softer approach

Peters' Verge piece links the announcement to election-season resistance: voters in several areas treated sprawling AI campuses as a ballot issue, not a distant tech story. When residents fear opaque deals, NDAs become a symbol of exclusion even if companies insist they are routine. Dropping them for public officials is therefore as much a trust signal as a legal tweak.

Amazon's October move to end NDAs with government agencies on data centers set a peer precedent. OpenAI's November commitments go further by bundling cost, jobs transparency and local-compact language. Tech-policy readers of Thai data centres coverage will see a pattern: hyperscale builders are racing to calm town-hall opposition before the next permitting fight.

What communities can demand next

Paying full electricity and infrastructure costs addresses a core complaint that AI loads raise bills for households that never asked for a campus nearby. Publishing job and hiring reports lets reporters and councils test whether promised employment materialises. Local compacts for owned compute campuses create a document trail for noise limits, water use, emergency planning and community benefits.

None of those pledges erase the physical footprint of large facilities. They do, however, change the negotiating baseline for readers who track AI infrastructure, including our Meta Conductor acquisition coverage. Officials can now refuse blanket secrecy and still protect narrowly defined secrets, while asking for cost coverage and public hiring data up front.

Verified reporting does not invent quotas, megawatt figures or site names beyond what OpenAI and The Verge disclosed. The durable news is the process change: fewer gag clauses for public officials, clearer cost responsibility, and formal community promises around jobs and locally tailored compacts. Jay Peters' story at The Verge is here: theverge.com.

Image: BalticServers data center. Wikimedia Commons (CC BY-SA 3.0).