I am not anti-AI. I think artificial intelligence could become one of the most useful inventions in human history. That is exactly why I do not want it excused from the standards we apply to everything else.
Safe AI is not a model that refuses to say something awkward. It is a system with boundaries, transparency and a clear line of accountability when it acts. It should be secure by design, private by default, and honest about what it is doing and why.
The harder question is authority. A system can be autonomous in operation without being sovereign in authority. It can complete a task, coordinate with other systems and improve over time while remaining subordinate to the people whose lives it affects. That distinction needs to be built into the architecture, not added to a terms-of-service page.
We should be wary of two failures at once: the failure to build useful things, and the failure to keep those things answerable. The first is a failure of ambition. The second is a failure of care.
WASH exists to argue for both. Build the technology. Open it where it should be open. Make it safe enough to trust with meaningful work. And never let capability become an excuse to stop asking who is responsible.