I thought privacy in AI was mostly a branding exercise. Every platform seemed to make the same promise and users were expected to trust policies they would never read.
Then I started noticing a different pattern around @OpenGradient and OpenGradient Chat. Instead of focusing on trust the system appears to focus on reducing what needs to be trusted in the first place. Messages are encrypted before leaving the device and identity is separated from the interaction before reaching the model.
What caught my attention isn't the technology itself but the incentive shift.
Build privacy into the system itself not just the marketing and you’ll see user behavior shift. The changes are subtle though.
So does lowering privacy friction create real new demand Or do most users still choose convenience and what they already know.
That's the part I don't think the market has answered yet.
For now I'm less interested in the claims and more interested in the behavior. I'm watching whether people start sharing information with AI that they previously held back and whether that changes how tools like OpenGradient Chat are used over time.