Why API Security Is Foundational to AI Security (Akamai Study)
Secure AI starts with securing the APIs underneath it. This session breaks down why APIs and MCPs are the control plane for AI apps and agents, where unmanaged API risk hides, and how security and platform teams can improve visibility, posture, runtime protection, and governance across production environments.
KEY TAKEAWAYS
- APIs and MCPs power AI agents that read data, invoke tools, and take actions, making API security foundational to AI security.
- AI-linked APIs, shadow and rogue APIs, and zombie or deprecated endpoints create blind spots that traditional governance often misses.
- Akamai's fourth annual API security impact study surveyed more than 1,800 security professionals across 10 countries and found that 87% reported an API security incident in the past 12 months.
- Top risks for APIs connected to LLMs include data leakage, endpoint exploitation, prompt injection through malicious inputs, and lack of visibility.
- Kong and Akamai can mirror production traffic out of band from Kong Gateway to analyze API requests and responses, identify sensitive data and posture issues, and support optional blocking without adding application latency.
TIMESTAMPS
00:00 - Intro
00:12 - Speaker introductions
00:56 - Risks behind AI apps
01:34 - AI agent sprawl
02:28 - APIs as control plane
03:31 - Unmanaged API blind spots
05:01 - API security impact study
06:38 - LLM API risks
08:38 - Incident rates rise
09:52 - Sensitive data visibility
10:38 - Governance gaps
12:15 - Security team implications
12:55 - Four security pillars
15:43 - Agentic AI security
16:52 - Kong integration workflow
18:47 - Key takeaways
19:32 - Next steps
LINKS
Explore Kong's API security solutions: https://konghq.com/solutions/api-security
Read more about API incident costs and AI threats: https://konghq.com/blog/enterprise/cost-of-api-security-incidents-2025
#APISecurity #AISecurity #AIGateway #AgenticAI #Kong