Amazon's decision to drop nondisclosure agreements in data center negotiations with local governments is a quiet but significant admission: secrecy was the problem, not the solution. Microsoft made the same move earlier this year, and together these shifts signal that the industry is finally recognizing what community opposition has been saying all along. When residents cannot learn what is being built in their backyard until after the deal is done, trust evaporates, and moratoriums follow. Hundreds of them, from New York to San Francisco, have already been proposed or enacted. The question now is whether transparency becomes a genuine operating principle or just another concession made after the damage was already done.
For our readers, this matters beyond the headlines about data centers. The same instinct that drove those NDAs, the belief that complexity justifies opacity, is alive in the broader push toward AI agents. As When Code Outpaces Reading, It's Time to Hire AI points out, we are already handing substantial decision-making power to systems that can process more code than any human could read. If we cannot trust the organizations deploying these tools to be transparent about basic infrastructure, how do we extend that trust to agents operating in our personal and professional lives? The parallel is direct: both cases hinge on whether the people affected get to see what is coming before it arrives.
There is also a lesson here about how innovation actually spreads. Amazon and Microsoft are not acting out of altruism; they are responding to a backlash that made secrecy costly. That is the same dynamic that drives progress in AI itself, as explored in Can AI Agents Pursue Discovery Without a Defined Goal. Open-ended exploration only works when the system is allowed to surface what it finds, even when that is inconvenient. Likewise, community trust only works when companies are willing to surface their plans early, even when that invites scrutiny. The firms that treat transparency as a constraint rather than a feature will keep generating resistance; the ones that build it into the process from the start will find that opposition often softens into negotiation.
The practical takeaway is simple: watch what happens next. Dropping NDAs is a policy change, but it does not guarantee meaningful engagement. The real test is whether Amazon and Microsoft publish their environmental impact assessments, share construction timelines, and open real channels for community input before shovels hit the ground. If they do, the moratorium wave may recede. If they do not, we will see the same fights, just with better press releases. For anyone building AI infrastructure, or relying on it, the message is the same: transparency is not a concession to critics. It is the only durable path to the permission society grants when it feels respected. And if the industry learns that lesson here, it might carry it into the far harder question of how much access we give AI agents to our data and our decisions.