Follow ZDNET: Add us as a favorite source On Google.
Key takeaways of zdnet
- Hackers found in a report can take advantage of an autorun feature in Karsar.
- The danger is “important”, but an easy fix.
- Cursor uses AI for assistance with code-editing.
A new report has revealed what he describes as “an important security vulnerability” in the cursor, the popular AI-managed code-editing platform.
ReportPublished on Wednesday by software company Oasis Security, it was found that the code repository within the cursor in which .vscode/tasks.json configuration can be instructed that as soon as repository is opened, instructions can be given to run some tasks automatically. Hackers can exploit the autorun feature that is embedded in the code via malware.
Also: I coded 24 days in 12 hours with $ 20 AI tool – but a big loss is
Oasis wrote, “It has the ability to leak sensitive credentials, modify files or work as a vector for comprehensive system compromise, keeping cursor users at significant risk from supply chain attacks,” Oasis wrote.
While Cloud Codes and Windsurf software developers have become popular among Karsar and other AI-operated coding tools, technology is still filled with bugs. Answer, another AI coding assistant who introduced its latest agent earlier this week, recently removed the entire database of a company.
Safety fault
According to Oasis report, the problem lies in the fact that the “Workplace Trust” facility of the cursor is disabled by default.
Originally, the purpose of this feature is to be a verification step for cursor users so that they only go to code that they know and believe. Without this, the platform would automatically run the code in a repository, which would leave the window open for bad actors to slip into malware for bad actors, which can then endanger the user’s system – and from there, potentially spread to a broad network.
Also: I asked AI to modify the Mission-Critical Code, and what happened next
The code running without the workplace trust facility can open a straight route for unauthorized access with an organization-wide blast radius, “Oasis said.
In a statement by Oasis published in the report, Karsar stated that its platform is defaultly operated with the workplace trust as it intervenes with some corps automatic features that users depend regularly.
The company said, “We either recommend using a basic text editor while enabling the workplace trust or working with suspected malicious repository,” the company said.
Also: He can put the new cloud feature ‘you can risk your data,’ anthropic admits
Karsar also told Oasis that he would soon publish an update security guideline about the workspace trust facility.
How to be preserved
The solution is to enable the workplace trust facility only in the cursor. To do this, add the following security prompt to settings, and then restart the program:
,
“Security.workspace.Trust.enabled”: Truth,
“Security.workspace.trust.startupprompt”: “Always”
ZDNET has reached Karsar for further comments.
