Skip to content

Image: isc.sans.edu · rights & removal

Executive Summary

A phishing attempt involved a message directing the recipient to a link hosted on Google Cloud Storage that included the recipient's email address in the URL fragment. Upon visiting this page, the recipient was presented with a generic login form personalized using their email domain and attempted to use a screenshot service to fetch a background image from the victim's organization's legitimate website. The source for the screenshot retrieval failed on one target but succeeded on another. After source code inspection, the page contained a credential-harvesting kit that collected passwords, public IP addresses, geographic location, browser information, and language. This information, along with a timestamp, was intended to be sent to an attacker via the Telegram Bot API. The collection process included performing DNS-over-HTTPS queries for MX records associated with the victim's email domain using Google's public DNS service. These retrieved MX records were then included in the data sent alongside the stolen credentials. This technique is noted as a method of gathering organizational infrastructure details during credential theft, even though the attacker could independently look up these records.

Facts Only

* A phishing page was linked from a message to an HTML page hosted on Google Cloud Storage.
* The recipient's email address was included in the URL fragment of the link: `hxxps[:]//storage[.]googleapis[.]com/alx2/03/1a/we.html#[recipiente-mailaddress]`.
* The page presented a generic login form personalized with the victim's email domain.
* The page attempted to use the thum.io service to retrieve a screenshot of the victim's organization's website as a background.
* The DNS-over-HTTPS query for MX records was performed using Google's public DNS service against the victim's email domain.
* The retrieved MX record results were included in the information sent to the attacker with stolen credentials.
* The credential harvesting kit collected the submitted password, visitor's public IP address, geographic location, browser information, and language.
* The collected data was transmitted to the attacker using the Telegram Bot API.
* A separate phishing page examined concurrently also collected a similar set of information and sent it via Telegram.
* Infoblox documented a phishing-as-a-service operation called Morphing Meerkat in March 2025, which used DNS-over-HTTPS queries for MX records.

Full Take

The scenario illustrates a sophisticated layering of reconnaissance during credential harvesting, moving beyond simple data exfiltration to infrastructure mapping. The inclusion of MX record lookups serves not just as an additive data point but as an automated method for the attacker to map the victim's entire email service provider ecosystem. While the specific phishing page did not use these records to customize the interface directly—unlike some services—the fact that the raw DNS query and result were bundled with credentials indicates a playbook focused on creating highly contextualized, high-value intelligence packets. This suggests an intent to provide the attacker with immediate context about where the compromised accounts reside, prioritizing subsequent attacks or negotiations. The observation that this technique mirrors documented phishing-as-a-service operations reinforces the idea that automated reconnaissance is becoming standard practice for threat actors, shifting the focus from purely transactional theft to comprehensive infrastructure profiling. The implication is that attackers are integrating passive reconnaissance into the very mechanism of data collection to maximize the utility of stolen credentials.
Bridge Questions:
How does the existence of readily available public DNS data change the risk assessment when an attacker integrates it into a phishing delivery mechanism? What countermeasures can be implemented against automated, multi-stage intelligence gathering embedded within credential flows? What is the long-term impact on organizational security awareness when attackers routinely include infrastructural data in malicious communications?

From the original · SANS Internet Storm Center

A phishing page interested in MX records I’ve mentioned before in some of my diaries that it can be instructive to look at what appear to be “generic” phishing messages in detail from time to time, since some of them might prove not to be as generic as they initially appear.
Read the full story at isc.sans.edu

Sentinel unavailable

The automated check of the source article's wording did not complete, so no result is shown.

This looks only at the wording of the original source article, not at this page's AI-written sections. A small local AI model made this estimate. It has not been checked against known human and machine texts, so treat it as provisional. It cannot show who wrote an article.

A phishing page interested in MX records, (Tue, Nov 17th) | Huntaegis