New macOS Infinity Stealer uses Nuitka Python payload and ClickFix


New macOS Infinity Stealer uses Nuitka Python payload and ClickFix

Pierluigi Paganini
March 30, 2026

Infinity Stealer targets macOS via fake Cloudflare CAPTCHA, using Nuitka; first such campaign per Malwarebytes.

Researchers at Malwarebytes spotted a new macOS infostealer, named Infinity Stealer, using a Python payload compiled with Nuitka. It spreads via ClickFix, tricking users with fake Cloudflare CAPTCHA pages.

“A fake verification page instructs the visitor to open Terminal, paste a command, and press Return. Once executed, the infection process begins immediately.” reads the report published by MalwareBytes. “The technique gained popularity on Windows systems, but it’s now being adapted for macOS, with the instructions tailored to the platform: Command + Space > open Terminal > paste the command”

The fake Cloudflare CAPTCHA tricks users into pasting a Terminal command that fetches a Stage-1 Bash dropper.

“The first payload is a Bash script using a template previously observed in macOS stealers such as MacSync (also referenced as SHub in earlier research).” continues the report. “This suggests the use of a shared builder.”

The dropper decodes the payload, writes the Stage‑2 binary, removes macOS protections, executes it, passes C2 data, and then deletes itself. Stage‑2 is a native macOS loader compiled with Nuitka that unpacks and runs the final Python stealer.

The final payload, UpdateHelper[.]bin, is a Python 3.11 stealer that collects browser credentials, Keychain entries, crypto wallets, .env files, and screenshots, exfiltrating data via HTTP. It detects analysis environments and adds random delays to evade detection. Once exfiltration finishes, it notifies the operator via Telegram and queues credentials for server-side cracking.

Infiniti Stealer shows macOS is no longer low‑risk, adapting Windows‑style ClickFix and using Nuitka to evade detection. If you ran suspicious Terminal commands, stop sensitive activity, change passwords from a clean device, revoke sessions and keys, and check /tmp and LaunchAgents. Experts recommend running a full scan with antimalware software. Never paste commands from websites, because no real CAPTCHA requires it.

The report includes Indicators of Compromise (IOCs) for this campaign.

Follow me on Twitter: @securityaffairs and Facebook and Mastodon

Pierluigi Paganini

(SecurityAffairs – hacking, newsletter)







Source link

Leave a Reply

Subscribe to Our Newsletter

Get our latest articles delivered straight to your inbox. No spam, we promise.

Recent Reviews


There’s something oddly brilliant about outsourcing your curiosity to an AI that doesn’t get tired or awkward. After all, if an AI agent can call thousands of pubs and build a Guinness price index, why stop there? Why not send one loose into the wild to track the cost of your daily caffeine fix or your late-night ramen cravings?

I’m sold — I want one of those

That’s exactly the kind of domino effect sparked by a recent experiment inspired by Rachel Duffy from The Traitors. A developer built an AI voice agent that sounded natural enough to chat up bartenders and casually ask for Guinness prices, compiling the data into a public index. It worked so well that most people on the other end didn’t even clock that they were speaking to a machine. And just like that, a slightly chaotic, very clever idea turned into something surprisingly useful.

Now imagine applying that same idea to coffee and ramen. Because if there are two things people are oddly loyal and sensitive about, it’s how much they’re paying for a flat white or a bowl of tonkotsu.

A “CaffIndex,” for instance, could map out the price of cappuccinos across cities, highlighting everything from overpriced aesthetic cafés to hidden gems that don’t charge $3 for foam. Similarly, a “Ramen Radar” could track where you’re getting the most bang for your broth, whether it’s a premium bowl or a spot that somehow gets everything right. Don’t giggle, I’m serious.

The appeal isn’t just novelty. It’s scale. Calling up a handful of places yourself is tedious. Getting real-time, city-wide data? Nearly impossible. But an AI agent doesn’t mind dialing a thousand numbers, repeating the same question, and logging every answer with monk-like patience. What you get in return is a living, breathing map of prices.

It’s not all sunshine and roses

Of course, it is not all smooth sipping and slurping. There is a slightly uneasy side to this, too. Questions around consent and transparency start to creep in, and you cannot help but wonder if every business would be okay with being surveyed by an AI that sounds just a little too real. In the original experiment, the AI was designed to be honest when asked directly, but let’s be real: most people aren’t going to question a friendly voice casually asking about prices. It feels harmless in the moment, and that is exactly what makes it a bit tricky.

Still, there is something genuinely exciting about the idea. Not in a scary, robots-are-taking-over kind of way, but in a way that makes you pause and think, this could actually be useful if handled right. Prices are creeping up everywhere, from your rent to that comforting bowl of ramen you treat yourself to after a long day. Having something that keeps track of it all feels like a small win.

Maybe that is the real takeaway here. Today it is Guinness. Tomorrow it could be your morning coffee or your go-to ramen spot. It makes you wonder how long it will be before your phone steps in, calls up a café, asks about their espresso, and saves you from spending more than you should. Because honestly, if AI is willing to do the boring work for you, the least it can do is make sure your next cup and your next bowl actually feel worth it.



Source link