AI Sec Watch

This survey paper reviews deep learning techniques for video inpainting (filling in missing or removed parts of videos) and methods to detect when videos have been inpainted. The research covers both how AI systems can reconstruct videos and how to identify when such reconstruction has occurred, addressing concerns about manipulated video content.

This research paper describes a defensive framework for cloud-native network slicing (dividing a network into isolated virtual segments) that uses multi-agent deep reinforcement learning (a type of AI where multiple learning agents work together to make decisions) to protect against security threats. The framework takes a proactive approach by continuously changing the network configuration to make it harder for attackers to find vulnerabilities, similar to a moving target that's difficult to hit.

This research paper presents a method for detecting living-off-the-land reverse shells (attacks where hackers use built-in system tools already present on a computer to create a backdoor connection back to their own machine) at a large scale by using data synthesis (artificially creating training examples rather than collecting real attack data). The approach aims to improve security detection systems' ability to identify these sneaky attacks that are hard to catch because they blend in with normal system activity.

Spring AI's MilvusVectorStore#doDelete(List) method has a vulnerability called filter-expression injection (where attackers can manipulate database filter commands by inserting malicious code into document IDs) because document IDs are not properly cleaned before being used. This affects Spring AI versions 1.0.0 through 1.0.x and 1.1.0 through 1.1.x.

This article covers week two of a lawsuit where Elon Musk is suing OpenAI and its leaders, claiming they broke promises to keep the company as a nonprofit dedicated to safe AI development. OpenAI's president Greg Brockman countered that Musk actually pushed for the company to become for-profit and wanted majority control, and that Musk is suing because he left the company in 2018 and now sees it as a competitor to his own AI company, xAI (an artificial intelligence system). Musk is seeking $134 billion in damages and wants to remove the current leadership and undo OpenAI's recent restructuring.

FastGPT, a platform for building AI agents, has a vulnerability in versions before 4.14.17 that allows attackers to send requests to internal or private network addresses without needing to log in. The problem is in the fetchData function, which retrieves data from user-provided URLs but doesn't properly check them against a blocklist (isInternalAddress) that's meant to prevent SSRF attacks (where a server is tricked into making requests to systems it shouldn't access).

FastGPT, a platform for building AI agents, had a security flaw in how it protected against SSRF attacks (server-side request forgery, where an attacker tricks a server into connecting to unauthorized internal systems). While some endpoints blocked internal network URLs, the tool creation endpoints did not, allowing an authenticated user to save a malicious internal URL that could later be used without additional checks when running workflows.

FastGPT, an AI platform for building AI agents, has a vulnerability in versions 4.14.11 and earlier where its isInternalAddress() function fails to block access to cloud metadata endpoints (services that store sensitive system information). Attackers can bypass the blocklist using URL encoding techniques (methods to disguise URLs), and because a security check is disabled by default, the metadata endpoint remains accessible without additional protection.

FastGPT versions 4.14.11 and earlier have a DNS rebinding vulnerability (TOCTOU, or Time-of-Check to Time-of-Use, where a check happens at one moment but the actual action uses a different result moments later) in their isInternalAddress() function. The function validates that a hostname resolves to a safe private IP address, but because the actual HTTP request performs a separate DNS lookup afterward, an attacker can change the DNS record between validation and the request, bypassing the security check.