AI Changed What You Ship. It Also Changed What You Have to Secure.
Mend.io, Monday, June 22nd, 2026
AI reshapes both the software teams ship and the security risks they must manage.
The article argues AI has fundamentally altered what software teams deliver and the security challenges they face.
Mend proposes categorizing risks around two objects: the software shipped, including AI-generated code, and the AI systems themselves.
AI-generated code introduces vulnerabilities at scale, with findings up 10x in six months and CVSS 7.0+ flaws appearing 2.5x more often than in human-written code.
AI systems face unique failure modes like prompt injection and jailbreaks requiring adversarial testing.
The solution is a single independent security layer that verifies both traditional software and AI components without bias.