Attackers turned Google ads and a Bing redirection mechanism into a trap for macOS users who were looking for Claude. The researchers found a scheme where the advertising result on request claude mac showed the real domain bing.com, but after several transitions led to a fake page of loading Claude.
The attack began with malicious advertising in Google search. After the click, the browser first passed through the usual Google advertising link, then through the service address bing.com/ck/a, which Bing takes into account the transitions from the search results. The next link was the real page of the South American Internet store, which, according to researchers, the intruders pre-hacked. From there, the visitor was sent to the false site Claude.
The long chain helped to hide the end page from automatic checks. The hacked site passed only visitors who came through Bing and fit the specified browser parameters. The fake page of Claude also checked where the site opened. During the direct transition, the visitor received an error of 404, so the specialist or automatic scanner could not see the trap at all.
On the page, the victim was offered to install Claude through the terminal. The site showed the real command of the Anthropic installation, but the copy button placed a completely different code in the clipboard. After the insertion, the team displayed the official claude.ai address in the terminal, in the background of the hidden link, downloaded the script from another server and transmitted it to the shell zsh for launch.
The reception refers to ClickFixClickFix: the attacker does not launch the malicious command himself, but convinces the user to copy and perform it under the guise of the usual instruction. In the described campaign, the bait was the installation of Claude, so the person actually ran the code with his own hands, thinking that he was installing an official client.
The researchers found several domains with the same design, the same command for macOS and a similar address structure. The authors of the investigation refer to the infrastructure as a set of AcSig tools that are monitored within the company. At the same time, it was not possible to determine the contents of the final downloaded file, so it is not yet possible to approve what kind of malware the victims received.
According to the researchers, about four of the five detected ClickFix attacks come to users through search engines. In the new campaign, criminals added another level of disguise: a person saw a familiar Bing domain already in the advertising result, and the malicious page was opened only after passing the entire prepared chain.
The attack began with malicious advertising in Google search. After the click, the browser first passed through the usual Google advertising link, then through the service address bing.com/ck/a, which Bing takes into account the transitions from the search results. The next link was the real page of the South American Internet store, which, according to researchers, the intruders pre-hacked. From there, the visitor was sent to the false site Claude.
The long chain helped to hide the end page from automatic checks. The hacked site passed only visitors who came through Bing and fit the specified browser parameters. The fake page of Claude also checked where the site opened. During the direct transition, the visitor received an error of 404, so the specialist or automatic scanner could not see the trap at all.
On the page, the victim was offered to install Claude through the terminal. The site showed the real command of the Anthropic installation, but the copy button placed a completely different code in the clipboard. After the insertion, the team displayed the official claude.ai address in the terminal, in the background of the hidden link, downloaded the script from another server and transmitted it to the shell zsh for launch.
The reception refers to ClickFixClickFix: the attacker does not launch the malicious command himself, but convinces the user to copy and perform it under the guise of the usual instruction. In the described campaign, the bait was the installation of Claude, so the person actually ran the code with his own hands, thinking that he was installing an official client.
The researchers found several domains with the same design, the same command for macOS and a similar address structure. The authors of the investigation refer to the infrastructure as a set of AcSig tools that are monitored within the company. At the same time, it was not possible to determine the contents of the final downloaded file, so it is not yet possible to approve what kind of malware the victims received.
According to the researchers, about four of the five detected ClickFix attacks come to users through search engines. In the new campaign, criminals added another level of disguise: a person saw a familiar Bing domain already in the advertising result, and the malicious page was opened only after passing the entire prepared chain.
