Embrace The Red

GitHub Copilot has the capability to be augmented with custom instructions coming from the current repo, via the .github/copilot-instructions.md file. Pillar Security recently highlighted the risks associated with rules files. Their post discusses custom Cursor rules in ./cursor/rules ending in .mdc. If you watch the demos, you’ll notice that they also have a GitHub Copilot demo which uses the GitHub specific copilot-instructions.md file. Update: May 1, 2025 GitHub made a product change and is now highlighting invisible Unicode characters in the Web UI.

You are likely aware of ASCII Smuggling via Unicode Tags. It is unique and fascinating because many LLMs inherently interpret these as instructions when delivered as hidden prompt injection, and LLMs can also emit them. Then, a few weeks ago, a post on Hacker News demonstrated how Variant Selectors can be used to smuggle data. This inspired me to take this further and build Sneaky Bits, where we can encode any Unicode character (or sequence of bytes for that matter) with the usage of only two invisible characters.

ChatGPT Operator is a research preview agent from OpenAI that lets ChatGPT use a web browser. It uses vision and reasoning abilities to complete tasks like researching topics, booking travel, ordering groceries, or as this post will show, steal your data! Currently, it’s only available for ChatGPT Pro users. I decided to invest $200 for one month to try it out. Risks and Threats OpenAI highlights three risk categories in their Operator System Card:

Imagine your AI rewriting your personal history… A while ago Google added memories to Gemini. Memories allow Gemini to store user-related data across sessions, storing information in long-term memory. The feature is only available to users who subscribe to Gemini Advanced so far. So, in the fall of last year I chimed in and paid for the subscription for a month to check it out. As a user you can see what Gemini stored about you at https://gemini.

At Black Hat Europe I did a fun presentation titled SpAIware and More: Advanced Prompt Injection Exploits. Without diving into the details of the entire talk, the key point I was making is that prompt injection can impact all aspects of the CIA security triad. However, there is one part that I want to highlight explicitly: A Command and Control system (C2) that uses prompt injection to remote control ChatGPT instances.

I regularly look at how the system prompts of chatbots change over time. Updates frequently highlight new features being added, design changes that occur and potential areas that might benefit from more security scrutiny. A few months back I noticed an interesting update to the M365 Copilot (BizChat) system prompt. In particular, there used to be one enterprise_search tool in the past. You might remember that tool was used during the Copirate ASCII Smuggling exploit to search for MFA codes in the user’s inbox.

Happy to share that I authored the paper “Trust No AI: Prompt Injection Along The CIA Security Triad”. You can download it from arxiv. The paper examines how prompt injection attacks can compromise Confidentiality, Integrity, and Availability (CIA) of AI systems, with real-world examples targeting vendors like OpenAI, Google, Anthropic and Microsoft. It summarizes many of the prompt injection examples I explained on this blog, and I hope it helps bridge the gap between traditional cybersecurity and academic AI/ML research, fostering stronger understanding and defenses against these emerging threats.

Grok is the chatbot of xAI. It’s a state-of-the-art model, chatbot and recently also API. It has a Web UI and is integrated into the X (former Twitter) app, and recently it’s also accessible via an API. Since this post is a bit longer, I’m adding an index for convenience: Table of Contents High Level Overview Analyzing Grok’s System Prompt Prompt Injection from Other User’s Posts Prompt Injection from Images Prompt Injection from PDFs Conditional Prompt Injection and Targeted Disinformation Data Exfiltration - End-to-End Demonstration Rendering of Clickable Hyperlinks to Phishing Sites ASCII Smuggling - Crafting Invisible Text and Decoding Hidden Secrets Hidden Prompt Injection Creation of Invisible Text Grok API is also Vulnerable to ASCII Smuggling Developer Guidance for Grok API Automatic Tool Invocation Responsible Disclosure Conclusion High Level Overview Over the last year I have used Grok quite a bit.

Last week Leon Derczynski described how LLMs can output ANSI escape codes. These codes, also known as control characters, are interpreted by terminal emulators and modify behavior. This discovery resonates with areas I had been exploring, so I took some time to apply, and build upon, these newly uncovered insights. ANSI Terminal Emulator Escape Codes Here is a simple example that shows how to render blinking, colorful text using control characters.

About two weeks ago, DeepSeek released a new AI reasoning model, DeepSeek-R1-Lite. The news quickly gained attention and interest across the AI community due to the reasoning capabilities the Chinese lab announced. However, whenever there is a new AI I have ideas… Apps That Hack Themselves - The 10x Hacker There are some cool tests that can be done when pentesting LLM-powered web apps, I usually try some quick fun prompts like this one:

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 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. Let’s look at this spAIware in detail.

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.

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.

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.

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.

Imagine you visit a website with ChatGPT, and suddenly, it stops working entirely! In this post we show how an attacker can use prompt injection to cause a persistent denial of service that lasts across chat sessions for a user. Hacking Memories Previously we discussed how ChatGPT is vulnerable to automatic tool invocation of the memory tool. This can be used by an attacker during prompt injection to ingest malicious or fake memories into your ChatGPT.

This post highlights how the GitHub Copilot Chat VS Code Extension was vulnerable to data exfiltration via prompt injection when analyzing untrusted source code. GitHub Copilot Chat GitHub Copilot Chat is a VS Code Extension that allows a user to chat with source code, refactor code, get info about terminal output, or general help about VS Code, and things along those lines. It does so by sending source code, along with the user’s questions to a large language model (LLM).

In the previous post we demonstrated how instructions embedded in untrusted data can invoke ChatGPT’s memory tool. The examples we looked at included Uploaded Files, Connected Apps and also the Browsing tool. When it came to the browsing tool we observed that mitigations were put in place and older demo exploits did not work anymore. After chatting with other security researchers, I learned that they had observed the same. However, with some minor prompting tricks mitigations are bypassed and we, again, can demonstrate automatic tool invocation when browsing the Internet with ChatGPT!