Click Here! Trust Me
Malvertising, as defined by Wikipedia, is the use of online advertising to spread malware. The name is a portmanteau of ‘malicious advertising’, and it relies on social engineering and trust to deliver its payload, much like phishing schemes. It used to be more obvious, to the point where it was meme-able, the province of sidebar ads on explicit websites luring in the unwary with hot singles in one’s area or for performance enhancing pharmaceuticals. It still exists in banner and clickbait type ads, commonly promoting self help objects or special ‘deals’ for products. It’s a regular feature of many preinstalled casual game suites in Windows; my home computer’s antivirus software has blocked the pages of those ads from opening when accidentally clicked nearly every single time due to the URL’s being suspicious. To a degree, it’s targeted. The ones I see are aimed at a particular demographic, usually older people who may not be as savvy in noticing the falseness.
The web seems to be riddled with fake ads. Popups, compromised widgets and third party apps are all likely locations to find either self hosted malware or redirects to spoofed domains. And these days, many of them are targeted towards developers. There are two articles regarding the delivery of malware via spoofed websites on my news feed today. One is a ‘classic’ ClickFix campaign, tricking users into downloading an open source video editor containing a REMCOS Trojan through a faked CAPTCHA page. Pretty standard fare, as ClickFix campaigns go. The other one is a Claude AI clone that delivers an infostealer as its payload, and is the one I’ll be focusing on.
Spoofing is frequently the vehicle driving these campaigns, presenting a legitimate looking site during a search, as is the case with the Claude clone, which has been dubbed ‘InstallFix’ by Push Security. The malicious code is typically obfuscated. Push Security’s report states that the fake site is identical in every way to the real thing, and is only visible as malware if one hovers over the copy and paste command. Hence my often repeated warning against copying and pasting anything into a command line. You don’t know what it actually is. AI is a popular vector for compromise, since the agents offer less technically adept users a simplified way to utilize developer tools and is trendy to boot.
Push goes on to say that the copy/paste/run model is becoming the norm for developer software. Many packages, in a variety of coding languages, instruct their users to do it as part and parcel of installation. It is a framework based on trusting the domain, which is where a lot of these problems share a root cause. Despite numerous admonitions towards zero trust, security is rarely at the forefront of a user’s mind when faced with something flashy or there’s a deadline to meet. And threat actors are counting on that lack of vigilance.
InstallFix is distributed through Google ads, showing up in sponsored results when users are searching for ‘Claude Code’, ‘Claude Code install’, or ‘Claude Code CLI’. In the images Push shares in the article, not one of the top results actually goes to Anthropic, Claude’s parent company, and the only place one should be getting it from.
This in itself is part of the larger issue with social engineering and the degree to which our search engines have changed. Sponsored results are how many of these malvertising scams work. People see the first link and click on it without thinking, or even looking at the source. I’ve come across it myself in everything from social media to banking institutions to my phone provider. A few extra seconds to check the source of a site is all it takes not to click the wrong thing. Prevention is always going to be the best medicine, but remember if you get into trouble, your friendly neighborhood WISP is here to help.
Posted, 3/10/26
