Fable 5 banned overnight: the lesson for SMBs
The US government made a major AI model vanish in 24 hours. What it reveals about the real risk when your business depends on a single model.
A major AI model, switched off in 24 hours
In mid-June 2026, Anthropic cut access to two of its models, Fable 5 and Mythos 5. Not an outage, not an update: a decision by the US government. According to NBC, Bloomberg and Al Jazeera, the administration ordered access suspended for any foreign national, on national security grounds. Rather than filter country by country, Anthropic ended up disabling the models for everyone to comply.
The stated reason? The government believed a "jailbreak" method existed. According to Anthropic, it amounted to a limited ability to review code and fix errors — a capability rival models also have. The company publicly pushed back, arguing a "narrow jailbreak" didn't justify pulling a model used by hundreds of millions of people. The order was enforced anyway.
The real issue isn't Fable 5
Whether you use Claude, GPT or something else, note the mechanism: a model that millions of people relied on disappeared in a day, with no useful warning, for a reason entirely outside its users' control.
For an SMB, that's a signal. If your customer support, invoicing or prospecting runs on , you're exposed to things you don't control: