October 9, 2026
Amazon told TechCrunch it will stop using non-disclosure agreements when negotiating data center deals with local governments, following a similar move from Microsoft earlier this year. Read that sentence again: the two biggest cloud landlords in the world are only now agreeing to let the communities hosting their power-hungry buildings know what's actually in the contracts. That's not generosity. It's damage control.
Secrecy around these deals has been fueling real backlash -- TechCrunch notes hundreds of proposed and enacted moratoriums on data center construction, from New York to San Francisco. When a town finds out after the fact that it quietly signed away tax breaks, water rights, or electricity capacity to a hyperscaler, the reaction isn't gratitude. It's a moratorium. Amazon and Microsoft are reading the room, but ending NDAs only tells residents what deal was struck -- it doesn't give them a seat at the table before it's struck, and it doesn't address the underlying tension: these facilities consume enormous amounts of power and water in communities that often don't benefit proportionally from the jobs or tax revenue promised.
For any business watching the AI infrastructure buildout from the outside, the lesson generalizes beyond data centers: transparency after the fact is cheap, and it's not the same as trust built in advance. We've made this point before about AI agents and the trust gap -- disclosure is a floor, not a ceiling. If you're a company evaluating AI vendors or infrastructure partners, the real question isn't whether they'll eventually tell you what happened, but whether their default posture is openness or concealment. Check out ViibeStack's own approach to this at our Trust Center, including how we handle incident response -- because when something goes wrong, waiting for a leak or a lawsuit to force disclosure is a bad look no matter how big you are.
Also in TechCrunch this week: MIT's Dr. Sherry Turkle revisits a theme she's studied for decades -- that humans can't help forming emotional bonds with things that merely act like they care, from chatbots to robot companions. Her point, as TechCrunch summarizes it, is that even the most primitive AI exchange triggers something in us that feels like being cared for, and we respond by caring back. It's a psychological fact, not a feel-good one.
This matters commercially in a way that's easy to miss. Every customer-facing AI chatbot, every AI assistant embedded in a helpdesk or CRM workflow, is quietly exploiting this wiring whether its designers intend to or not. A customer who feels "cared for" by a support bot is more forgiving, more loyal, and more likely to disclose information they'd otherwise withhold -- which is useful, but also a responsibility. If your AI tooling leans on that emotional illusion to retain users or extract data, you own the ethics of that choice, not just the UX win. Turkle's warning isn't that AI companionship is inherently bad; it's that we're building it faster than we're thinking about what we owe the humans who respond to it as if it were real.
Put Amazon's NDA reversal next to Turkle's research and you get the same story told twice. In both cases, powerful AI-adjacent systems -- data centers and conversational agents -- were built and deployed at scale before the public had a real framework for evaluating what they were agreeing to. Communities signed infrastructure deals blind. Users formed attachments to software blind. The industry is now backfilling transparency after the fact, which is better than nothing, but it keeps putting the burden of catching up on everyone else.
If you're a business leader choosing AI vendors right now, take both stories as a prompt to ask harder questions upfront rather than trusting a retroactive policy change: what does this vendor disclose by default, and how does their product actually behave with the people who use it day to day?
Would you trust a tech company's transparency pledge more if it came before a deal was signed, or is after-the-fact disclosure good enough for you?
Sources