GitHub Adds 2 Third-Party AI Agents as 83% of Developers Use AI but Only 7% Fully Trust It
Updated
Updated · InfoWorld · Oct 13
GitHub Adds 2 Third-Party AI Agents as 83% of Developers Use AI but Only 7% Fully Trust It
1 articles · Updated · InfoWorld · Oct 13
Summary
GitHub has put third-party coding agents including Claude and Codex into public preview, letting prompts or issues turn into proposed changes, pull requests or follow-up review tasks on its platform.
The move targets a widening trust gap: Stack Overflow says AI use in development rose from 44% in 2023 to 83% in 2026, while only 7% of developers trust AI for important work decisions without verification.
GitHub says those agents get the same security protections and limits as Copilot cloud agents, giving teams a consistent place to inspect, test and approve AI-generated code even if they switch model providers.
That consistency matters because 48% of developers say they trust AI only when output is easy to verify, making governance, review rules and regression checks more valuable as more agents propose code.
GitHub is also testing optional Copilot approvals for pull requests, but its own guidance says AI reviews can miss problems, underscoring that enterprises still need human oversight for sensitive changes.