A Top AI Model Went Dark for 6 Days
Anthropic's newest model went offline worldwide for 6 days, then returned paid. The lesson on building so you can swap models.
Last week a top AI model went dark worldwide for 6 days, and the companies building on it had no say.
Anthropic’s newest model went offline everywhere from June 12 to June 18 under a US export order, then came back as a paid feature. Teams that were running on it one week were locked out the next, with no warning they could act on.
This is the risk that never makes it into the pilot plan. You build a workflow around one model from one vendor, and a policy you do not control can pull it out from under you overnight.
The model you pick matters less than whether your process survives it changing. So build to allow you can swap. Train on the fundamentals. Keep your data clean and your process documented, and the model becomes a part you can replace.
We sell chemicals, and we manage data, and the data is the part you actually own.
If a vendor change ever broke a workflow on you, what did it cost to fix it?