LIVE ALERT
⚠️ DailySamchar.in सूचना: सर्वर मैंटेनेंस कार्य 11 तारीख को दोपहर 2:00 PM से 3:20 PM तक रहेगा। इस दौरान वेबसाइट बंद रहेगी। असुविधा के लिए खेद है। || Planned Maintenance: Server will be down on 11th Sep from 02:00 PM to 03:20 PM. We apologize for the inconvenience.

Trojan Horse in the Search Results: Fake Claude AI Sites Hijack Google and Bing to Deliver Malware

Trojan Horse in the Search Results: Fake Claude AI Sites Hijack Google and Bing to Deliver Malware

How a fake AI download scam starts with Google

The attack begins with a standard Google search for “Claude Mac.” According to security researchers at Push Security, a sponsored search result appears, often displaying bing.com as its destination. While that address might seem reassuring because Bing belongs to Microsoft, clicking the advertisement initiates a series of redirects. First, Google’s advertising system processes the click. Next, Bing’s legitimate click-tracking service forwards the browser to a compromised WordPress website previously owned by a legitimate retailer. Finally, that website redirects the visitor to a fraudulent Claude download page.

The destination website, claude-desk-code.com, mimics the official Claude installation experience. By the time a user arrives at the final landing page, they have been guided through a chain of legitimate services, making the journey appear safe. Researchers note that the attackers deliberately used a real Bing search result for a compromised website to leverage Bing’s click-tracking redirect, effectively cloaking their true destination behind a trusted domain.

Why the fake Claude installer is difficult to spot

The scammers employ a technique known as “cloaking” to evade detection by security scanners. The compromised retailer’s website runs internal checks to verify whether visitors arrived through Bing and if their browsers share specific metadata. Similarly, the fake Claude website checks if visitors originated from Google or Bing. If a security researcher or an automated bot visits the malicious website directly, the site may serve a 404 error page instead of the fake installer.

This mechanism allows the attackers to hide their malicious content from unwanted scrutiny while presenting the deceptive installer only to targeted victims. By observing these defensive maneuvers, researchers confirmed the existence of at least two separate checks designed to protect the infrastructure from outside investigation. This layered approach ensures that the malicious site remains active longer, as it effectively filters out most automated security crawlers.

The fake Claude download button hides a dangerous command

The most deceptive feature of the campaign involves a copy-to-clipboard button. The website displays an installation command that appears to originate from Anthropic, the developer of Claude. This text looks like standard command-line instructions that a proficient Mac user might expect. However, when the user clicks the “Copy” button, the page secretly places a completely different command onto the system clipboard.

The substituted command utilizes Base64 encoding to obfuscate an attacker-controlled website. When a user pastes this into the Mac Terminal, the initial output may display a fake, reassuring message indicating that Claude is being downloaded from an official source. In the background, the command decodes the hidden address, downloads a remote script, and pipes it directly into the Mac’s Z shell for execution. This process effectively tricks the user into authorizing the download of external, potentially malicious code while believing they are simply installing a legitimate AI assistant.

What happens if someone runs the fake Claude installer?

The malicious command is designed to fetch a script from a remote server and execute it with the privileges of the active user. While researchers have not confirmed the final payload, the potential impact of such an attack is significant. Running an untrusted Terminal command grants the script the ability to interact with the system, modify files, or install persistence mechanisms, such as launch agents that run every time the computer boots.

The risk is amplified by the fact that the victim performs the action manually. By disguising the instructions as a necessary installation step, the scammers exploit the user’s trust in the software provider. Data from Push Security suggests that approximately 80% of “ClickFix” style attacks identified by their team involve search engine manipulation, highlighting a growing trend where attackers use the credibility of major platforms like Google and Bing to bypass traditional user caution.

Protecting your Mac from sophisticated search scams

Avoiding these threats requires a combination of behavioral changes and technical vigilance. To minimize risk, users should avoid clicking on sponsored search results for software downloads. Instead, navigate directly to the developer’s official website—in this case, anthropic.com—or use official app stores where available.

When encountering websites that request the use of Terminal commands, users should exercise extreme skepticism. If a site provides a “Copy” button, paste the content into a plain-text editor before executing it. This allows for an inspection of the code, where one might spot encoded strings, suspicious web addresses, or scripts that execute immediately upon input. Furthermore, maintaining an updated version of macOS and utilizing reputable, real-time antivirus software provides a necessary layer of defense. If a device has already been exposed to such a command, immediate steps include disconnecting from the internet, performing a thorough security scan, and updating credentials for all sensitive accounts via a secure, separate device.

Disclaimer: This content is auto-generated for informational purposes only.

Source: Read Original News

Leave a Reply

Your email address will not be published. Required fields are marked *