2026-09-11 · Fuente：TechRadar
New data highlighted by TechRadar points to a widening gap between how much enterprises trust agentic AI code and how much they can actually see, verify and govern.
Enterprises are extending growing trust to agentic AI code at the same time as their actual visibility into what that code does fails to keep pace, according to data highlighted in a recent TechRadar analysis. The result is a widening visibility gap: a divergence between how much organizations believe they can rely on autonomously generated and executed code and how much they can actually see, verify and govern.
The central finding is directional: the gap is widening rather than holding steady. As agentic systems take on more of the writing, modification and deployment of code, enterprise confidence has grown faster than the tooling and processes needed to track what that code actually contains. Risk, in the framing of the report, is effectively smuggled into codebases — not through a single dramatic breach, but through the quiet accumulation of code that no one has fully examined. Trust, in other words, is running ahead of oversight.
For organizations buying and operating AI compute, the finding lands on the software layer that sits directly on top of accelerated infrastructure. Agentic code is increasingly the code that gets shipped, and if confidence in it outstrips visibility into it, governance becomes the weak link in the stack. Procurement and platform decisions typically assume that what runs in production has been reviewed; a widening trust-versus-visibility gap erodes exactly that assumption.
The implication is that visibility cannot remain an afterthought. If enterprises continue to extend trust to agentic code faster than they build the ability to inspect it, the gap becomes a standing channel for risk to enter production environments unnoticed. Closing it means treating oversight of autonomously generated code as a precondition for trusting it — measuring what agentic systems actually write and do before confidence hardens into complacency.
Etiquetas: Agentic AI · AI Code Security · Enterprise AI Governance · Code Visibility · Software Supply Chain Risk
萬安算交所 AIXX · AIXX · Exchange verificado · Ir al mercado