Krux
April 10, 2026
Gartner: One in Four AI Apps Will Hit Five Security Breaches by 2028
Published: April 10, 2026 at 12:42 AM
Updated: April 10, 2026 at 12:42 AM
100-word summary
Gartner predicts 25% of enterprise AI applications will suffer at least five minor security incidents annually by 2028, nearly triple the 9% rate in 2025. Major incidents will jump from 3% to 15% by 2029. The culprit? Model Context Protocol, the tool letting AI agents talk to each other and your data, was designed for speed and flexibility, not security. MCP ships with no default security enforcement. Gartner warns that any AI touching sensitive data while ingesting outside content and communicating externally is a "no-go zone." Companies need authentication built for agents, not borrowed from human login systems. The chatbot you deployed last month might need permission controls tighter than...
What happened
Gartner predicts 25% of enterprise AI applications will suffer at least five minor security incidents annually by 2028, nearly triple the 9% rate in 2025. Major incidents will jump from 3% to 15% by 2029. The culprit? Model Context Protocol, the tool letting AI agents talk to each other and your data, was designed for speed and flexibility, not security. MCP ships with no default security enforcement. Gartner warns that any AI touching sensitive data while ingesting outside content and communicating externally is a "no-go zone." Companies need authentication built for agents, not borrowed from human login systems.
Why it matters
The chatbot you deployed last month might need permission controls tighter than your intern's.