Most Organisations Face an Unsecured API Surge As AI Agents Outpace Security

Salt Security's 1H 2026 State of AI and API Security report reveals a critical gap between AI agent deployment and security readiness. 92% of organisations lack advanced security maturity for agentic environments, while 66% saw API growth exceeding 50% in the past year. Nearly all (99%) attack attempts originate from authenticated sources, with rogue AI agents exploiting legitimate credentials. 65% of attacks exploit security misconfigurations (OWASP API8), amplified by over-permissioned APIs connected to AI agents. Salt Security proposes an 'Agentic Security Graph' model to map relationships between LLMs, MCP servers, and APIs as a unified security discipline.