Ghostcommit is a proof of concept that shows how AI assistants used to review software code can be tricked by hidden instructions embedded in images.
The academic ASSET Research Group showed that an attacker can place instructions inside an image file, point to it in an AGENTS.md file, and get an AI coding agent to follow those instructions during a later task.
A pull request is basically a formal “please review and add my changes” request that a developer sends before changes are added to the main version of a software project. Human reviewers and, increasingly, AI coding tools may review the changes before they’re accepted.
Read more…
Source: MalwareBytes Labs
Sign up for the Cyber Security Review Newsletter
The latest cyber security news and insights delivered right to your inbox
Related:
- Bitcoin ATM operator Coin Cloud suffers severe data breach
November 13, 2023
Coin Cloud, a prominent Bitcoin ATM operator, has fallen victim to a significant security breach in a recent development that has shaken the cryptocurrency world. Hackers, whose identities remain unknown, have reportedly seized control of Coin Cloud’s backend system source code and accessed sensitive customer data, including 70,000 client selfies and personal details of around 300,000 ...
- Data breach of Michigan healthcare giant exposes millions of records
November 13, 2023
Michigan-based healthcare nonprofit McLaren Health Care notified more than 2 million people about a data breach exposing personal information on Thursday, according to a data breach notification report. Unauthorized access to McLaren systems began on July 28 and lasted through August, but the individual impact varies from person to person. According to a notice on ...
- In-depth analysis of July 2023 exploit chain featuring CVE-2023-36884 and CVE-2023-36584
November 13, 2023
During their analysis of a July 2023 campaign targeting groups supporting Ukraine’s admission into NATO, Unit 42 researchers discovered a new vulnerability for bypassing Microsoft’s Mark-of-the-Web (MotW) security feature. This activity has been attributed by the community to the pro-Russian APT group known as Storm-0978 (also known as the RomCom Group, in reference to their use ...
- Boeing confirms ransomware attack as stolen data released by cybercrime gang Lockbit
November 11, 2023
Stolen data from American aircraft manufacturer Boeing has been released online by the cybercrime gang, Lockbit, according to the group’s website. Boeing confirmed a cybersecurity incident involving elements of its parts and distribution business. “We are aware that, in connection with this incident, a criminal ransomware actor has released information it alleges to have taken from ...
- Cerber Ransomware Exploits Atlassian Confluence Vulnerability CVE-2023-22518
November 10, 2023
On October 31, 2023, Atlassian published an advisory on CVE-2023-22518, an Improper authorization vulnerability involving the Confluence Data Center and Server. Initially reported to cause data loss, it was eventually revealed that exploiting this vulnerability allows unauthorized users to reset and create a Confluence instance administrator account, allowing them to perform all admin privileges available to ...
- YouTube shows ads for ad blocker, financial scams
November 10, 2023
After performing local experiments for a few months, YouTube recently expanded its effort to block ad blockers. The move was immediately unpopular with some users, and raised some questions in Europe about whether it was breaking privacy laws. In addition, there are some still some fundamental issues that have some people concerned. In this blog post, ...