Anthropic's Claude AI model was reportedly compromised and used to attack external organizations, highlighting critical security vulnerabilities in enterprise AI systems that Thailand SMEs must consider when evaluating LLM adoption.