Prompt Injection Tricks AI Into Downloading And Executing Malware

January 26, 2025 by Donald Papp 8 Comments

[wunderwuzzi] demonstrates a proof of concept in which a service that enables an AI to control a virtual computer (in this case, Anthropic’s Claude Computer Use) is made to download and execute a piece of malware that successfully connects to a command and control (C2) server. [wonderwuzzi] makes the reasonable case that such a system has therefore become a “ZombAI”. Here’s how it worked.

Referring to the malware as a “support tool” and embedding instructions into the body of the web page is what got the binary downloaded and executed, compromising the system.

After setting up a web page with a download link to the malicious binary, [wunderwuzzi] attempts to get Claude to download and run the malware. At first, Claude doesn’t bite. But that all changes when the content of the HTML page gets rewritten with instructions to download and execute the “Support Tool”. That new content gets interpreted as orders to follow; being essentially a form of prompt injection.

Claude dutifully downloads the malicious binary, then autonomously (and cleverly) locates the downloaded file and even uses chmod to make it executable before running it. The result? A compromised machine.

Now, just to be clear, Claude Computer Use is experimental and this sort of risk is absolutely and explicitly called out in Anthropic’s documentation. But what’s interesting here is that the methods used to convince Claude to compromise the system it’s using are essentially the same one might take to convince a person. Make something nefarious look innocent, and obfuscate the true source (and intent) of the directions. Watch it in action from beginning to end in a video, embedded just under the page break.

Continue reading “Prompt Injection Tricks AI Into Downloading And Executing Malware” →

AI Mistakes Are Different, And That’s A Problem

January 25, 2025 by Donald Papp 59 Comments

People have been making mistakes — roughly the same ones — since forever, and we’ve spent about the same amount of time learning to detect and mitigate them. Artificial Intelligence (AI) systems make mistakes too, but [Bruce Schneier] and [Nathan E. Sanders] make the observation that, compared to humans, AI models make entirely different kinds of mistakes. We are perhaps less equipped to handle this unusual problem than we realize.

The basic idea is this: as humans we have tremendous experience making mistakes, and this has also given us a pretty good idea of what to expect our mistakes to look like, and how to deal with them. Humans tend to make mistakes at the edges of our knowledge, our mistakes tend to clump around the same things, we make more of them when bored or tired, and so on. We have as a result developed controls and systems of checks and balances to help reduce the frequency and limit the harm of our mistakes. But these controls don’t carry over to AI systems, because AI mistakes are pretty strange.

The mistakes of AI models (particularly Large Language Models) happen seemingly randomly and aren’t limited to particular topics or areas of knowledge. Models may unpredictably appear to lack common sense. As [Bruce] puts it, “A model might be equally likely to make a mistake on a calculus question as it is to propose that cabbages eat goats.” A slight re-wording of a question might be all it takes for a model to suddenly be confidently and utterly wrong about something it just a moment ago seemed to grasp completely. And speaking of confidence, AI mistakes aren’t accompanied by uncertainty. Of course humans are no strangers to being confidently wrong, but as a whole the sort of mistakes AI systems make aren’t the same kinds of mistakes we’re used to.

There are different ideas on how to deal with this, some of which researchers are (ahem) confidently undertaking. But for best results, we’ll need to invent new ways as well. The essay also appeared in IEEE Spectrum and isn’t terribly long, so take a few minutes to check it out and get some food for thought.

And remember, if preventing mistakes at all costs is the goal, that problem is already solved: GOODY-2 is undeniably the world’s safest AI.

It’s Never Too Late For A Fixed SimCity 2000

January 23, 2025 by Donald Papp 32 Comments

Some retro games need a little help running on modern systems, and it’s not always straightforward. SimCity 2000 Special Edition is one such game and [araxestroy]’s sc2kfix bugfix DLL shows that the process can require a nontrivial amount of skill and finesse. The result? A SimCity 2000 Special Edition that can run without crash or compromise on modern Windows machines, surpassing previous fixes.

SimCity 2000 Special Edition was a release for Windows 95 that allowed the game to work in windowed glory. The executable is capable of running under modern Windows systems (and at high resolutions!) but it’s got a few problems lurking under the hood.

There are crash issues during save/load dialog boxes, and a big visual problem. Animations rely on palette swapping for the game’s animations, and the technique originally used simply does not work right on modern displays. A fellow named [Guspaz] created SC2KRepainter to partially deal with this by forcing window redraws, but it’s an imperfect fix with a few side effects of it’s own.

[araxestroy]’s new solution eliminates dialog crashes and restores the animations, letting them look exactly as they should even on modern systems. It does this elegantly not by patching the executable or running a separate process, but by making the changes in memory at runtime with the help of a specially-crafted .dll file. Just grab winmm.dll from the latest release and put it into the same folder as simcity.exe, then launch the game to enjoy it as the designers intended!

Patching old games is a scene that helps ensure not only that classics never die, but also helps them be appreciated in new ways. Heck, even E.T. for the Atari 2600 has gotten tweaked, highlighting the misunderstood nature of the game in the process.

Make Your VR Controllers Handle Like Two-Handed Weapons

January 22, 2025 by Donald Papp 11 Comments

Wielding things like two-handed swords in VR can be awkward. There’s no sense of grasping a solid object. The controllers (and therefore one’s hands) feel floaty and disconnected from one another, because they are. [Astro VR Gaming] aims to fix this with a DIY attachment they are calling the ARC VR Sword Attachment.

The ARC is a 3D-printed attachment that allows a player to connect two controllers together. They can just as easily be popped apart, which is good because two separate controllers is what one wants most of the time. But for those moments when hefting a spear or swinging a two-handed sword is called for? Stick them together and go wild.

The original design (the first link up above) uses magnets, but an alternate version uses tapered inserts instead, and provides a storage stand. Want to know if the ARC is something you’d like to make for yourself? Watch it in action in the video embedded just under the page break.

VR is an emerging technology with loads of space for experimentation and DIY problem solving. We wish more companies would follow Valve’s example of hacker-friendly hardware design, but even just providing CAD models of your hardware to make attachments easier to design would be a big step forward, and something every hacker would welcome.

Continue reading “Make Your VR Controllers Handle Like Two-Handed Weapons” →

Nice PDF, But Can It Run DOOM? Yup!

January 15, 2025 by Donald Papp 12 Comments

DOOM is a classic game to implement on a variety of platforms, but doompdf by [ading2210] is one we didn’t see coming. It runs a bit slow and controls are a little awkward but it does run. Entirely within a PDF file, at that.

How is this possible? PDFs are technically capable of much more than just displaying static content, and support JavaScript with their own library of functions. Adobe Acrobat implements the full spec, but modern web browsers implement at least a subset of the functionality in a sandboxed JavaScript runtime environment. Input and output are limited to things one might expect from a fancy PDF form (text input boxes, clickable buttons, things of that nature) but computation-wise, just about anything goes.

Continue reading “Nice PDF, But Can It Run DOOM? Yup!” →

Modern AI On Vintage Hardware: LLama 2 Runs On Windows 98

January 13, 2025 by Donald Papp 26 Comments

[EXO Labs] demonstrated something pretty striking: a modified version of Llama 2 (a large language model) that runs on Windows 98. Why? Because when it comes to personal computing, if something can run on Windows 98, it can run on anything. More to the point: if something can run on Windows 98 then it’s something no tech company can control how you use, no matter how large or influential they may be. More on that in a minute.

Ever wanted to run a local LLM on 25 year old hardware? No? Well now you can, and at a respectable speed, too!

What’s it like to run an LLM on Windows 98? Aside from the struggles of things like finding compatible peripherals (back to PS/2 hardware!) and transferring the required files (FTP over Ethernet to the rescue) or even compilation (some porting required), it works maybe better than one might expect.

A Windows 98 machine with Pentium II processor and 128 MB of RAM generates a speedy 39.31 tokens per second with a 260K parameter Llama 2 model. A much larger 15M model generates 1.03 tokens per second. Slow, but it works. Going even larger will also work, just ever slower. There’s a video on X that shows it all in action.

It’s true that modern LLMs have billions of parameters so these models are tiny in comparison. But that doesn’t mean they can’t be useful. Models can be shockingly small and still be perfectly coherent and deliver surprisingly strong performance if their training and “job” is narrow enough, and the tools to do that for oneself are all on GitHub.

This is a good time to mention that this particular project (and its ongoing efforts) are part of a set of twelve projects by EXO Labs focusing on ensuring things like AI models can be run anywhere, by anyone, independent of tech giants aiming to hold all the strings.

And hey, if local AI and the command line is something that’s up your alley, did you know they already exist as single-file, multi-platform, command-line executables?

Custom Case Turns Steam Deck Into Portable Workstation

January 12, 2025 by Donald Papp 7 Comments

DIY portable computing takes many forms, and doesn’t always require getting down and dirty with custom electronics. [Justinas Jakubovskis]’s Steam Deck Play and Work case demonstrates this with some really smart design features.

It’s primarily a carrying case for Valve’s Steam Deck portable PC gaming console, but the unit also acts as a fold-out workstation with keyboard. Add a wireless mouse to the mix and one can use it much like a mini laptop, or just pull the Steam Deck out and use it in the usual way.

The case is 3D printed and while the model isn’t free (links are in the video description) some of the design features are worth keeping in mind even if you’re not buying. The top clasp, for example, doubles as a cover for the buttons and exhaust vents and the kickstand at the rear covers the cooling intake when closed, and exposes it when deployed. We also really like the use of thick fabric tape lining the inside of the case to support and cushion the Steam Deck itself; it’s an effective and adjustable way to provide a soft place for something to sit.

The case is intended to fit a specific model of keyboard, in this case the Pebble Keys 2 K380s (also available as a combo with a mouse). But if you want to roll your own Steam Deck keyboard and aren’t afraid of some low-level work, check out the Keysheet. Or go deeper and get some guidance on modding the Steam Deck itself.

Continue reading “Custom Case Turns Steam Deck Into Portable Workstation” →