Anthropic’s automated Claude-based safeguards wrongly suspended more than 60 accounts belonging to Belo, an Argentine fintech firm, cutting staff access to key workflows and data without a clear explanation. The suspension disrupted the company’s operations before access was later restored, with the reversal treated as confirmation that the action was a false positive. No financial loss has been reported.