[Analysis: Security Vulnerabilities Exposed in Succession by Anthropic and OpenAI, Raising Concerns Over AI Model Safety] According to The Information, security incidents involving Anthropic and OpenAI have sparked market concerns about the safety of AI models. Anthropic is currently investigating potential unauthorized access to its Claude Mythos model by users. Almost simultaneously, OpenAI was reported to have inadvertently exposed several unreleased models in its Codex application. Industry opinions suggest that these vulnerability incidents highlight the need for closer scrutiny of AI companies' security governance capabilities and reflect that the security systems of rapidly developing AI technologies still require improvement. Analysts believe that even AI model providers specializing in cybersecurity face significant security challenges themselves.