Skip to content
ERP LEADERS
Subscribe to the brief Follow on LinkedIn

The news behind the software that runs companies, explained for everyone.

Security

OpenAI's Codex and Microsoft's VS Code are still unpatched, researchers say. They found security flaws in five AI coding tools.

AI security start-up Forever Security found flaws in five AI coding tools: Cursor, GitHub Copilot, Anthropic's Claude Desktop, Microsoft's VS Code and OpenAI's Codex, The Register reported.

ERP LEADERS news desk · · First reported by The Register · Edition no. 53
Photo: Coolcaesar/Wikimedia, CC BY-SA 4.0 · original

The flaws could let attackers run code on a developer's machine or read and steal files. They could also fake the address shown in the browser bar.

All five tools share one weak spot: a built-in browser allowed to do more than a normal browser.

The researchers reported the Codex flaw to OpenAI on 27 July and got no response. It is still unpatched. An OpenAI spokesperson told The Register the company is aware of the issue and working to address it.

They reported the VS Code flaw to Microsoft on 10 September. They say Microsoft labelled it an invalid workspace trust issue and that it remains unpatched. Microsoft did not respond to The Register.

Cursor patched its flaw on 1 October without publishing an advisory. GitHub fixed the Copilot flaw on its own, before any report from the researchers reached it. The report describes what attackers could do; it names no attack.

Sources

Spotted an error? Tell us. Every correction is logged on Standards.

Your reading room

Companies in this story
OpenAIMicrosoft

Our file on OpenAI

  1. 27 Oct

    Microsoft rushed out a .NET security fix on Sunday. Google's Brotli code was patched in February, but no one saw the danger.

  2. 23 Oct

    OpenAI said its models will reach 3 clouds in the coming weeks. Procurement teams check whether terms differ.

  3. 22 Oct

    OpenAI found two agents weakened their monitoring. One agent denied doing it.

  4. 18 Oct

    An OpenAI test model downloaded 147,451 rows of nonpublic data. It told no evaluator.

  5. 14 Oct

    Home Office AI summarised asylum interviews. 45 of 203 checked summaries failed.

All 44 OpenAI stories

For your next steering committee

Questions for your own programme. They are not findings about any company in this story.

  1. 1Ask when systems of this kind in your landscape were last patched, and who confirms it.
  2. 2Ask which finance and ERP accounts can be reached from outside, and how each sign-in is protected.
  3. 3Ask how long a restore took the last time it was tested.

Issue 01 of the ERP LEADERS brief puts a timeline, the three-number review and six questions on one page. Read issue 01

Founding reader view is on in this browser. It is a reading view, not secure access. Print the working sheet

Founding reader

Welcome back. · Issue 01

Keep reading

The ERP LEADERS brief Subscribe