Anthropic banned 31 of our accounts overnight

Last week, Anthropic banned 31 Speed and Function from Claude — 23 team accounts and 8 personal accounts linked to the same credit card. No warning. No explanation. We filed appeals, contacted support — zero response from Anthropic. Even their chatbot was inaccessible.

Then it got stranger: Anthropic sent emails saying our accounts were restored. They weren’t. We’re still blocked.

We spent days and thousands of dollars making this invisible to our clients — individual workaround accounts, extra staff hours, emergency bridging. All because one vendor decided to cut us off overnight.

We’re not alone. I researched this for 2 days and found 210+ similar cases in the last month:

• An Argentine fintech (belo) had 60+ accounts shut down overnight (Patricio Molina Manuel Beaudroit Edwin Rager)

• A 110-person agricultural company lost access with zero explanation

• A university lab got banned for using the same payment card

• Adults flagged as children for mentioning school topics

Anthropic banned 1.45M accounts in H2 2025. Only 3.3% of appeals succeeded. And the only companies that got fast resolution were those that went viral on social media.

We evaluated every alternative — OpenAI, Gemini, Perplexity. None match Claude for what we need: non-technical team members running agentic automations from their desktops. That’s what makes this so frustrating — we can’t easily leave, and they won’t let us stay.

Full research: https://lnkd.in/drHJ7TEW

The moral: If your business depends on a single AI provider, you don’t own your workflow — they do. And they can take it away in one email with a link to a Google Form.

Three things we’re doing now:

1) Building a multi-model abstraction layer — Claude, GPT, Gemini interchangeable

2) Never letting one vendor control critical client delivery again

3) Treating AI providers like cloud infrastructure — always have a failover

Has this happened to your team?

#Anthropic

#Claude

#AI

#LessonsLearned