Can you trust your chatbot? Inside three AI-powered cyberattacks

This IBM Security Intelligence podcast episode examines three categories of AI-powered cyberattacks: prompt injection and poisoning of enterprise chatbots, LLM-assisted malware and exploit development, and coordinated AI agent swarms deployed against organizational infrastructure. The central finding is that AI is lowering the technical barrier for attackers while simultaneously expanding the attack surface enterprises expose through chatbot and AI agent deployments. Defenders must account for new threat vectors introduced by AI systems themselves—not just AI-assisted versions of legacy attacks. The episode frames AI chatbot trustworthiness as a material enterprise security governance question, with implications for how organizations design controls around AI agent autonomy and data access.