Anthropic's pre-IPO shares have declined following the US government's shutdown of its most powerful AI model, Fable 5, due to security concerns. The government cited a method to bypass the model, which Anthropic argues is a narrow and previously known vulnerability. The company contends that similar vulnerabilities exist in other models like OpenAI's GPT-5.5 without requiring a bypass. Anthropic, known for its safety-first AI approach, disputes the government's directive, claiming the evidence provided is insufficient. The company plans to release more details about the alleged jailbreak within 24 hours. The shutdown has negatively impacted Anthropic's IPO prospects, with market reactions reflecting uncertainty over whether the government's order will be reversed or extended to other AI models.