Microsoft's AI-driven Bing Chat responses are currently being tainted with harmful ads, which promote counterfeit download platforms distributing malware. Bing Chat, powered by OpenAI's GPT-4 engine, was launched by Microsoft in February 2023 to rival Google's supremacy in the search engine industry.
Bing Chat offers users an interactive experience that is more chat based, rather than the traditional search question and result format. The aim of this is to make the process more intuitive and user-friendly.
In March, Microsoft initiated the insertion of advertisements into Bing Chat conversations as a means to generate revenue through this innovative platform.
Nonetheless, the integration of ads into Bing Chat has inadvertently created an avenue for threat actors, who are increasingly employing search advertisements to disseminate malware.
Moreover, the conversational nature of interactions with AI-powered chat tools can foster an undue sense of trust, potentially leading users to click on ads—a behaviour less common when browsing impersonal search results.
This dialogue-driven interaction has the potential to confer a misplaced aura of authority and reliability upon URLs provided by AI, exacerbating the pre-existing issue of advertising within search platforms due to the introduction of AI assistants.
The labelling of these ads as 'promoted results' when a user hovers over a link in Bing Chat conversations may be insufficient in mitigating the associated risks.
Mimicking a widely-used IP scanner
Malicious advertisements are masquerading as download sources for the renowned 'Advanced IP Scanner' tool. This tool has previously been utilised by the RomCom RAT and Somnia ransomware operators.
Researchers observed that when inquiring about how to download Advanced IP Scanner via Bing Chat, the chat interface would provide a link for downloading the software.
However, upon hovering over a hyperlinked text within the chat, Bing Chat might initially display an advertisement, followed by the genuine download link. In this instance, the sponsored link turned out to be an advertisement that propagated malware.
The campaign was created by someone who hacked into the account of a legitimate Australian business to create two harmful ads that were targeting system admins and lawyers.
When users click on the malevolent advertisement associated with the IP scanner, they are directed to a website ('mynetfoldersip[.]cfd'). This site employs checks on IP addresses, time zones, and various system markers to distinguish between automated bots and human targets, particularly scrutinising for indications of sandbox or virtual machine environments.
Subsequently, the victims are redirected to 'advenced-ip-scanner[.]com,' a deceptive clone of Advanced IP Scanner that utilises typosquatting to deceive visitors, as evidenced by the intentional misspelling with an extra 'e' in 'advenced.'
Within the downloaded MSI installer, there are three files, including a heavily obfuscated malicious script. This script is designed to establish a connection with an external resource for the purpose of retrieving the payload.
The ultimate payload for this malware campaign was unable to be located, leaving us uncertain about the specific malware being ultimately deployed.
Nevertheless, in analogous campaigns, threat actors frequently disseminate malware designed for information theft or remote access, facilitating breaches into other accounts or corporate networks.
The presence of malvertising within Bing Chat conversations underscores the growing landscape of cyber threats, underscoring the importance of user caution when interacting with chatbots and the need to consistently verify URLs before downloading any content.
