Docker says developer trust in AI agents is the gap sandboxes must close

What the survey found about trust in AI

In a blog post published on Oct. 16, 2026, Docker's Terri Avnaim pointed to the latest Stack Overflow Developer Survey. According to the post, the survey was released earlier that month and reached more than 30,000 respondents across 169 countries. Docker describes it as a de facto standard for understanding how developers think and act, after 16 years of annual editions.

The post highlights several results on AI:

• Only 7% of developers trust AI output for important work decisions.
• 48% trust it only when they can verify it themselves.
• Security and privacy ranked as the second biggest factor in whether teams adopt or expand AI tools, behind only accuracy.
• Nearly one in four developers said AI-related security or privacy issues cost their team valuable time in the past year.

Docker reads these numbers as a gap between what developers want to do with agents and what they can securely and confidently run in production.

Containers still matter, but sandboxes are the focus

The company also noted its standing in the same survey. Docker is the most used (59%) and most desired (38%) tool in the survey's cloud development category. Docker presents those figures as a reflection of trust built over more than a decade of helping developers make software portable, open, reproducible and secure.

At the same time, Docker says it is putting its weight behind sandboxes. The post recalls that Docker president Mark Cavage devoted his opening keynote at WeAreDevelopers World Congress North America to the difference between sandboxes and containers. Using live demos, he argued that sandboxes, and not containers, provide the level of isolation needed to contain autonomous agents. Docker stresses that this does not mean it thinks less of containers, which it expects to keep playing a vital role in software development.

New products and what comes next

According to the post, Docker launched Docker Cloud Sandboxes at WeAreDevelopers and published the next-generation Sandbox Kit specification. The company says it has committed to submitting the specification to the Cloud Native Computing Foundation. Docker says nearly everything it showed was built around one idea: helping developers "manufacture trust for AI agents."

The post describes this as the start of a transformative era for Docker and promises more announcements in the coming weeks and months. It closes with the line that Docker "will continue to run on trust."

For developers, the practical offer is Docker Sandboxes. Docker says users can start an agent on their laptop, ship longer jobs off to the cloud, and bring the results back. The company offered a $250 credit to those who signed up before Oct. 31.

The argument is a vendor's own, and the survey figures are presented through Docker's lens. Still, the numbers it cites point to a clear pattern: most developers want to verify AI output themselves, and security concerns weigh heavily when teams decide whether to expand their use of AI tools. Docker is betting that stronger isolation for agents will be part of the answer.