Anthropic blocks entire company out of Claude without a reasonable explanation

Anthropic froze access to Claude for the entire company because of what it cited as "breaking the service's usage policy", without being any clear and only offering a Google Form to report the issue. The incident happened last week when more than 60 employees of Belo were locked out of the AI tool, Toms Hardware reported. Since they relied heavily on Claude for daily workflows, it created a major hassle for the company. They still do not know exactly which rules were broken.

Belo CEO Patricio Molina, wrote on X, "@claudeai you took down our entire organization with 60+ accounts belonging to a legitimate company for no apparent reason, without any explanations...Very bad UX and customer service." He wrote that Anthropic did not explain before it cut off access. An automated email supposedly only stated that the staff could not use Claude, without revealing exactly what had happened. It stated that "the automated systems detected a high volume of signals associated with your account which violate our Usage Policy."

With no access to Claude, the company was caught off guard as the employees did not know what to do. Their work relied heavily on Claude and the AI assistant's integrations, skills, and conversation histories. As the news spread across other companies, the creators faced a lot of backlash. The outrage led to Anthropic restoring access after 15 hours. However, it isn't clear whether it did so because of the criticism or whether it realised something had gone wrong on its end.

After Belo got the service back, the AI provider said that it happened because of a false positive, possibly because of some automated system that Anthropic uses. Several people flocked to Patricio Molina's X post and said that they had been facing similar problems for months. Many users claimed they had been filling out these Google Forms, but have not received any response from Anthropic. The responses show that the company penalises immediately, while never clarifying what went wrong.

While Patricio repents relying entirely on one tool, users commented that he should not have gone all in with Claude. They suggested other methods, such as running his own models locally, like OpenClaw. However, that involves running your own infrastructure, but still, people are saying that companies need to be safe rather than sorry, as was seen in Belo's case.

Relying on a single AI provider creates a significant supply chain bottleneck, and to avoid problems, experts recommend a "multi-model" strategy where companies should maintain active integrations with at least two independent AI services. Even if this leads to some operational overlap or duplicated costs, it serves as vital insurance against service outages, sudden policy shifts, or platform deprecation.