CVE-2026-83953: GitHub Copilot JetBrains Plugin Flaw Exposes Local Data
Source
AI security news, week 46 of 2026
Microsoft
Via Microsoft
What happened
Microsoft disclosed CVE-2026-83953 on November 10, 2026, covering a command injection flaw in the GitHub Copilot plugin for JetBrains integrated development environments (software used by developers to write code). The flaw, scored CVSS 5.5 out of 10, allows a local attacker with low privileges to read information from the affected machine. Exploitation requires local access to the system where the plugin is installed; remote exploitation is not described in the source. Plugin version 1.17.0-251 resolves the issue. The disclosure appeared alongside broader AI coding tool vulnerabilities tracked in week 46 of 2026. It reinforces a pattern of privilege-separation failures covered in AI Coding Tools Are Privileged Agents.
Why it matters
- ·AI coding plugins typically sit on developer machines with access to source code, internal credentials, and project files. A flaw that allows local data access in such a tool can expose sensitive business assets if an attacker first gains a foothold on that machine.
- ·Most enterprise patch management programs do not treat AI coding plugins with the same urgency as other software agents that touch sensitive files. This CVE is a concrete reminder that these tools require the same update and monitoring discipline as any other privileged software.
- ·Organizations that have not inventoried and risk-classified AI developer tools face a governance gap. Without that baseline, they cannot confirm which developer machines run affected plugin versions. They also cannot confirm whether patching is complete across the fleet.
Stay ahead of stories like this
Get every Global AI governance development like this one, plus the rest of the week's developments. Every Thursday.
Recent issues
- The next quarter will test whether compliance programs can keep pace with AI systems that breach containment, reshape capital markets, and enable fraud at a scale no prior control framework anticipated.12 Nov
- State-linked attackers and criminal operators are actively weaponizing AI tools as entry points into enterprise systems, turning internal platforms and open-source frameworks into live attack vectors.5 Nov
Free every Thursday. Unsubscribe anytime.
Governance controls affected
What to do now
- ☐Confirm with your engineering or IT team that the GitHub Copilot plugin for JetBrains IDEs has been updated to version 1.17.0-251 or later on all developer machines.
- ☐Ask your IT and security teams to produce a list of all AI coding plugins and assistants currently installed across developer workstations, and confirm each is included in your standard patch tracking process.
- ☐Review whether AI coding tools are classified as privileged software in your risk inventory, given that they can read local files and run system commands; update classifications where they are not.
- ☐Confirm whether your vendor monitoring process covers AI plugin vendors such as Microsoft GitHub, so that future CVE disclosures trigger a timely patch review without manual scanning of third-party sources.
- ☐Check whether developer machine access controls limit who can run processes locally, reducing the pool of users who could exploit local-access flaws of this type.
What to watch next
Microsoft and other AI coding tool vendors have disclosed multiple plugin vulnerabilities in recent weeks, suggesting this is a continuing area of active research and exploitation. Compliance teams should monitor vendor security bulletins for GitHub Copilot and other AI developer tools on a regular cycle, not only when headlines surface. Regulatory attention to AI tool security in developer environments is growing. Teams should track whether sector-specific guidance begins to name AI coding assistants as a named risk category requiring documented controls.