Palo Alto Unit42 discovered a vulnerability in the Google Cloud Vertex AI software development kit (SDK) for Python, and responsibly disclosed it to Google. Before Google’s fix, the vulnerability would have allowed an attacker operating entirely from their own Google Cloud project to hijack a victim’s model upload and poison it. By exploiting this flaw in vulnerable versions of the SDK, an attacker can achieve remote code execution (RCE) within a target’s Vertex AI serving infrastructure, with zero initial access to the victim’s project.
Read more…
Source: Palo Alto Unit 42
Sign up for the Cyber Security Review Newsletter
The latest cyber security news and insights delivered right to your inbox
Related:
- FBI warns about snoopy smart TVs spying on you
December 3, 2019
She laughed. I laughed. The TV laughed. I shot the TV. “Blasted Decepticons!” That’s how a popular meme went after the Transformer movies hit it big. Today, it’s not so funny. A recent FBI report warned smart TV users that hackers can also take control of your unsecured TV. “At the low end of the risk spectrum, they can ...
- Android: New StrandHogg vulnerability is being exploited in the wild
December 2, 2019
Security researchers from Promon, a Norwegian firm specialized in in-app security protections, said they identified a bug in the Android operating system that lets malicious apps hijack legitimate app, and perform malicious operations on their behalf. In a comprehensive report published today, the research team said the vulnerability can be used to trick users into granting intrusive permissions ...
- Exploit code published for dangerous Apache Solr remote code execution flaw
November 25, 2019
Confusion still surrounds a security bug that the Apache Solr team patched over the summer, which turns out it’s actually much more dangerous than anyone thought. Apache Solr is a Java-based open-source search engine, initially developed to add search functionality to the CNET website. The project was donated to the Apache Software Foundation in 2006, from where ...
- Critical Flaws in VNC Threaten Industrial Environments
November 22, 2019
The open-source Virtual Network Computing (VNC) project, often found in industrial environments, is plagued with 37 different memory-corruption vulnerabilities – many of which are critical in severity and some of which could result in remote code execution (RCE). According to researchers at Kaspersky, they potentially affect 600,000 web-accessible servers in systems that use the code. The ...
- High-Severity Windows UAC Flaw Enables Privilege Escalation
November 20, 2019
Researchers disclosed details of a high-severity Microsoft Windows vulnerability that could give attackers elevated privileges – ultimately allowing them to install programs, and view, change or delete data. The bug stems from User Account Control (UAC), a security feature of Windows within Secure Desktop which helps prevent unauthorized changes to the operating system. “With UAC fully ...
- New Roboto botnet emerges targeting Linux servers running Webmin
November 20, 2019
A cybercrime group is enslaving Linux servers running vulnerable Webmin apps into a new botnet that security researchers are currently tracking under the name of Roboto. The botnet’s appearance dates back to this summer and is linked to the disclosure of a major security flaw in a web app installed on more than 215,000 servers — ...