Cisco’s 2025 State of AI Security Report offers an in-depth analysis of AI-related security trends, threats, and policy developments. It examines rising risks such as infrastructure compromise, prompt injections, and data poisoning, alongside the misuse of AI by threat actors for social engineering and influence operations. The report also explores challenges posed by agentic AI and vulnerabilities from model fine-tuning and jailbreaking. On the policy side, it reviews U.S. state laws, global regulations like the EU AI Act, and international AI safety initiatives. The report concludes with risk management recommendations and highlights Cisco’s AI Defense solutions for securing enterprise AI systems.
