Private AI: what if your AI model goes offline tomorrow?
An AI model you rely on daily is suddenly unreachable. Not an outage, but a government directive from above. Anthropic pulled Mythos 5 and Fable 5 offline worldwide on orders from the US government, and it touches every organisation that has put AI into its processes. The question is not which model is best, but how you keep control with private, stable AI when a model like that disappears.
A block that has nothing to do with you still hits you
On 12 June 2026 <a href="https://www.anthropic.com" target="_blank" rel="noopener noreferrer" class="text-wabber-400 hover:text-wabber-300 underline decoration-wabber-400/30 hover:decoration-wabber-300 transition-colors">Anthropic</a> pulled Mythos 5 and Fable 5 offline worldwide on orders from the US government. No outage, no technical problem, but a government decision on national-security grounds that no user had any say in. Fable 5 worked well, and we think it is a shame it was the one to go. At the same time we understand the call: a government that sees a security risk steps in. But an organisation that has built AI into its processes has to keep running stably, regardless of a decision like that. Read <a href="https://www.anthropic.com/news/fable-mythos-access" target="_blank" rel="noopener noreferrer" class="text-wabber-400 hover:text-wabber-300 underline decoration-wabber-400/30 hover:decoration-wabber-300 transition-colors">Anthropic's official statement</a>.
The question is not how good the model is
The conversation about AI is almost always about performance, speed and new possibilities. That is understandable, because that is where the difference shows. Just as important is the dependency you build on a party you have no say over. A model that impresses today is gone tomorrow if the vendor or a government decides so. A government stepping in is the extreme case. The ordinary case is everyday: vendors change terms, prices and available regions whenever it suits them. This block only makes visible what was always true. The question is not how good the model is, but who decides whether you get to keep using it.
Availability is a business risk, not an IT detail
As long as AI is a standalone tool someone opens now and then, downtime is manageable. But the moment AI sits inside your processes, availability becomes a business risk. Our AI runs in the system itself: data comes in, gets processed and enriched, and flows on to the next step. If that model goes down, it is not one employee who stops, but a process. We know what that means, because our own systems run in production every day on our own infrastructure, not in a test setup. For an organisation that leans on AI, that is the difference between an inconvenience and an outage that reaches customers.
Sound familiar? Let us take a look.
Get in touch →This is why we build on private AI
For sensitive and critical systems we therefore look to private AI. Not because external models are bad, but because we would rather build on technology we manage ourselves than on technology we have to hope stays available. Our cluster runs on our own hardware in the Netherlands: 128GB VRAM, on-premise with you or hosted by us, no data and no availability that depends on a foreign party. ISO 27001 certification expected Q4 2026. You decide whether the model stays switched on. Whether you turn to us or to another party: do not let critical systems hang on a switch someone else can flip.
Marketing Coordinator at Wabber B.V.