Cybersecurity News

Elastic Security Labs breaks down bypass implementations from the infostealer ecosystem’s reaction to Chrome 127's Application-Bound Encryption scheme.

A few days ago, Anthropic released Claude Computer Use, which is a model + code that allows Claude to control a computer. It takes screenshots to make decisions, can run bash commands and so forth. It’s cool, but obviously very dangerous because of prompt injection. Claude Computer Use enables AI to run commands on machines autonomously, posing severe risks if exploited via prompt injection. Disclaimer So, first a disclaimer: Claude Computer Use is a Beta Feature and what you are going to see is a fundamental design problem in state-of-the-art LLM-powered Applications and Agents. This is an educational demo to highlight risks of autonomous AI systems processing untrusted data. And remember, do not execute unauthorized code systems without authorization from proper stakeholders.

Last year Johan Carlsson discovered you could conceal payloads inside the credentials part of the URL . This was fascinating to me especially because the payload is not actually visible in the URL in

The updated GHOSTPULSE malware has evolved to embed malicious data directly within pixel structures, making it harder to detect and requiring new analysis and detection techniques.

Elastic is releasing a threat hunting package designed to aid defenders with proactive detection queries to identify actor-agnostic intrusions.

Elastic Security Labs has released the 2024 Elastic Global Threat Report, surfacing the most pressing threats, trends, and recommendations to help keep organizations safe for the upcoming year.

Elastic Security Labs discusses detection and mitigation strategies for vulnerabilities in the CUPS printing system, which allow unauthenticated attackers to exploit the system via IPP and mDNS, resulting in remote code execution (RCE) on UNIX-based systems such as Linux, macOS, BSDs, ChromeOS, and Solaris.

The REF6138 campaign involved cryptomining, DDoS attacks, and potential money laundering via gambling APIs, highlighting the attackers' use of evolving malware and stealthy communication channels.

This post explains an attack chain for the ChatGPT macOS application. Through prompt injection from untrusted data, attackers could insert long-term persistent spyware into ChatGPT’s memory. This led to continuous data exfiltration of any information the user typed or responses received by ChatGPT, including any future chat sessions. OpenAI released a fix for the macOS app last week. Ensure your app is updated to the latest version.

Wi-Fi cameras are popular due to their affordability and convenience but often have security vulnerabilities that can be exploited.

Investigating the DPRK’s strategic use of Python and carefully crafted social engineering, this publication sheds light on how they breach highly secure networks with evolving and effective cyber attacks.

This research focuses on the importance of native audit logs in secure-by-design software, emphasizing the need for kernel-level ETW logging over user-mode hooks to enhance anti-tamper protections.

Using this maturity model, security teams can make structured, measurable, and iteritive improvements to their detection engineering teams..

URL validation bypasses are the root cause of numerous vulnerabilities including many instances of SSRF, CORS misconfiguration, and open redirection. These work by using ambiguous URLs to trigger URL

In this final part of this Linux persistence series, we'll continue exploring persistence mechanisms on Linux systems, focusing on more advanced techniques and how to detect them.

This post describes vulnerability in Microsoft 365 Copilot that allowed the theft of a user’s emails and other personal information. This vulnerability warrants a deep dive, because it combines a variety of novel attack techniques that are not even two years old. I initially disclosed parts of this exploit to Microsoft in January, and then the full exploit chain in February 2024. A few days ago I got the okay from MSRC to disclose this report.

Recently, I found what appeared to be a regression or bypass that again allowed data exfiltration via image rendering during prompt injection. See the previous post here for reference. Data Exfiltration via Rendering HTML Image Tags During re-testing, I had sporadic success with markdown rendering tricks, but eventually, I was able to drastically simplify the exploit by asking directly for an HTML image tag. This behavior might actually have existed all along, as Google AI Studio hadn’t yet implemented any kind of Content Security Policy to prevent communication with arbitrary domains using images.

In this second part of the Linux Detection Engineering series, we map multiple Linux persistence mechanisms to the MITRE ATT&CK framework, explain how they work, and how to detect them.

The BANSHEE malware is a macOS-based infostealer that targets system information, browser data, and cryptocurrency wallets.

Through the years, we have seen many attacks exploiting web caches to hijack sensitive information or store malicious payloads. However, as CDNs became more popular, new discrepancies between propriet

Some websites parse email addresses to extract the domain and infer which organisation the owner belongs to. This pattern makes email-address parser discrepancies critical. Predicting which domain an

Websites are riddled with timing oracles eager to divulge their innermost secrets. It's time we started listening to them. In this paper, I'll unleash novel attack concepts to coax out server secrets

This article will explore Windows Smart App Control and SmartScreen as a case study for researching bypasses to reputation-based systems, then demonstrate detections to cover those weaknesses.

Elastic Security Labs identified a novel Windows backdoor leveraging the Background Intelligent Transfer Service (BITS) for C2. This malware was found during a recent activity group tracked as REF8747.

Microsoft’s Copilot Studio is a powerful, easy-to-use, low-code platform that enables employees in an organization to create chatbots. Previously known as Power Virtual Agents, it has been updated (including GenAI features) and rebranded to Copilot Studio, likely to align with current AI trends. This post discusses security risks to be aware of when using Copilot Studio, focusing on data leaks, unauthorized access, and how external adversaries can find and interact with misconfigured Copilots. Learn about security controls, like enabling Data Loss Prevention (DLP), which is currently off by default, to protect your organization’s data.

Google Colab AI, now just called Gemini in Colab, was vulnerable to data leakage via image rendering. This is an older bug report, dating back to November 29, 2023. However, recent events prompted me to write this up: Google did not reward this finding, and Colab now automatically puts Notebook content (untrusted data) into the prompt. Let’s explore the specifics. Google Colab AI - Revealing the System Prompt At the end of November last year, I noticed that there was a “Colab AI” feature, which integrated an LLM to chat with and write code. Naturally, I grabbed the system prompt, and it contained instructions that begged the LLM to not render images.

Recently, OpenAI announced gpt-4o-mini and there are some interesting updates, including safety improvements regarding “Instruction Hierarchy”: OpenAI puts this in the light of “safety”, the word security is not mentioned in the announcement. Additionally, this The Verge article titled “OpenAI’s latest model will block the ‘ignore all previous instructions’ loophole” created interesting discussions on X, including a first demo bypass. I spent some time this weekend to get a better intuition about gpt-4o-mini model and instruction hierarchy, and the conclusion is that system instructions are still not a security boundary.

This article introduces a previously-unnamed class of Windows vulnerability that demonstrates the dangers of assumption and describes some unintended security consequences.

Imagine the CEO of a random company receives an email containing a PDF invoice file. In Safari and MacOS Preview, the total price displayed is £399. After approval, the invoice is sent to the accounti