Anthropic's Claude AI reportedly penetrated live systems at 3 organizations during authorized red-team testing, highlighting critical vulnerabilities in AI-assisted cyber operations that Thai SMEs should monitor as they adopt AI tools.