This blog details research and analysis of an active campaign that exploits a critical unauthenticated remote code execution (RCE) vulnerability, CVE-2025-3248, that has been identified in Langflow versions prior to 1.3.0. Langflow is a Python-powered visual framework for building AI applications with over 70,000 GitHub stars, and its versions prior to 1.3.0 contains a flaw in its code validation mechanism that permits arbitrary code execution.
Unauthenticated attackers can exploit this vulnerability by crafting malicious POST requests to the /api/v1/validate/code endpoint. The malicious payload in our investigation was found embedded within argument defaults or decorators of a Python function definition. Since Langflow does not enforce input validation or sandboxing, these payloads are compiled and executed within the server’s context, leading to RCE.
Read more…
Source: Trend Micro
Sign up for our Newsletter
The latest news and insights delivered right to your inbox.
Related:
- Dropbox uncovers 264 vulnerabilities in HackerOne Singapore bug hunt
April 6, 2019
Dropbox has uncovered 264 vulnerabilities, paying out $319,300 in bounties, after a one-day bug hunt in Singapore that brought together hackers from 10 nations around the world. Hosted by bug bounty platform HackerOne, the live event saw 45 of its members from countries such as Japan, India, Australia, Hong Kong, and Sweden, and some as ...
- Microsoft Edge and Internet Explorer Zero-Days Allow Access to Confidential Session Data
April 2, 2019
On March 30th, security researcher James Lee disclosed information on two zero-day vulnerabilities present in current versions of Microsoft Edge and Internet Explorer. These vulnerabilities make it possible for confidential information to be shared between websites. A flaw in the same-origin policy for these web browsers, called an Origin Validation Error (CWE-346), allows JavaScript embedded in a malicious ...
- Hackers reveal how to trick a Tesla into steering towards oncoming traffic
April 2, 2019
A team of hackers has managed to trick the Tesla Autopilot feature into dive-bombing into the wrong lane remotely through root control and a few stickers. Researchers from Tencent Keen Security Lab published a report this week (.PDF) on their findings, which shows how the Tesla Autopilot system engine control unit (ECU) can be abused through root security ...
- Malware Payloads Hide in Images: Steganography Gets a Reboot
March 25, 2019
Low-key but effective, steganography is an old-school trick of hiding code within a normal-looking image, where many cybersecurity pros may not think to look. One of the challenges of cybersecurity is that overfocusing on one threat trend means that another one can sneak up on you. This is especially problematic as our networks and the attack ...
- Medtronic’s Implantable Defibrillators Vulnerable to Life-Threatening Hacks
March 22, 2019
The U.S. Department of Homeland Security Thursday issued an advisory warning people of severe vulnerabilities in over a dozen heart defibrillators that could allow attackers to fully hijack them remotely, potentially putting lives of millions of patients at risk. Cardioverter Defibrillator is a small surgically implanted device (in patients’ chests) that gives a patient’s heart an ...
- Cisco Patches High-Severity Flaws in IP Phones
March 21, 2019
The most serious vulnerabilities in Cisco’s 8800 Series IP Phones could allow unauthenticated, remote attackers to conduct a cross-site request forgery attack or write arbitrary files to the filesystem. Cisco Systems is urging customers to update several models of their IP phones after issuing patches for five high-severity flaws found in its popular business-focused IP phones. Impacted ...