Cybercriminals Weigh Options for Using LLMs: Buy, Build, or Break?

While some cybercriminals have bypassed guardrails to force legitimate AI models to turn bad, building their own malicious chatbot platforms and making use of open source models are a greater threat.