Hackers are impersonating OpenAI, SpaceX, and Anthropic with fake event invitations designed to steal Microsoft and Google account credentials, authentication tokens, and session cookies. According to a report from Push Security, the campaign combines two phishing-as-a-service platforms—Kali365 and VoidProxy—to use separate attacks against Microsoft and Google accounts.
The attackers direct victims to convincing event-registration pages that promise free access to premium AI services and other benefits. The pages require visitors to sign in with a corporate account, allowing the criminals to capture credentials or authenticated sessions.
Push Security said it detected and blocked access to a malicious registration page in late October 2026. The campaign is notable because the same phishing page connects to both Kali365 and VoidProxy, using different authentication techniques depending on whether a victim selects Microsoft or Google.
Fake AI events lead to account phishing
It is unclear how the attackers initially distributed the phishing links. Push Security found that they used links from trusted services, including Google and Zapier URL shorteners, to redirect victims to malicious event pages.
Earlier versions of the campaign were distributed through the GoTo Webinar email system or official Zoom event pages. Similar activity was documented by ANY.RUN in July, when attackers used legitimate Zoom event pages to promote fake Meta, OpenAI, and Anthropic events that triggered device-code and adversary-in-the-middle (AiTM) phishing attacks.
Cofense also reported a phishing email that posed as an invitation to an Anthropic event. The email was distributed through GoTo Webinar’s infrastructure and redirected recipients to a fake registration page through multiple websites.
According to Push Security, two custom phishing kits tracked as ZoomMock and MSDRoot create the fake event pages and direct victims to the Kali365 and VoidProxy services. Researchers believe the same attackers operate both kits because they share infrastructure and target the same organizations.
The campaign’s pages are designed to resemble legitimate event websites hosted by OpenAI, SpaceX, and Anthropic. One page advertises an OpenAI event titled “ChatGPT for Social & Digital Marketing Teams” and promises participants three months of ChatGPT Pro.
Source: BleepingComputer
Another page promotes a private SpaceX event scheduled for Nov. 12 and offers registrants either three months of Grok Heavy or a “SpaceX share.” Researchers also found a page impersonating the now-deleted AWS re:Invent 2026 Anthropic event.
The fake registration pages tell visitors to sign in with their work account to complete registration and receive the advertised benefits. Before showing the page, the site displays a Cloudflare CAPTCHA claiming to verify that the visitor is not a robot. This can help prevent automated security platforms from scanning the phishing site.
Visitors are presented with two choices: “Continue with Microsoft” and “Continue with Google.”
Microsoft device-code phishing
When a visitor selects the Microsoft option, the page connects to Kali365, which generates a Microsoft device authorization code.
Source: BleepingComputer
Device-code phishing abuses the legitimate OAuth 2.0 device authorization grant flow to obtain an authentication token for a victim’s Microsoft Entra account. That token can provide access to the user’s resources and connected single sign-on applications, including Microsoft 365, Salesforce, Google Workspace, Dropbox, Adobe, SAP, Slack, Zendesk, Atlassian, and others.
Device-code attacks have traditionally been used for data theft and cyberespionage. In this campaign, victims are instructed to copy the code and enter it on Microsoft’s official device-login website, where they authenticate and approve the request.
Because the victim completes the process on Microsoft’s legitimate website, Kali365 can obtain the resulting Microsoft authentication token without directly asking the victim to reveal a password.
After the device-code authentication succeeds, the page redirects the victim to a VoidProxy Microsoft login page. This second stage can allow the attacker to collect the victim’s password and session cookies as well.
Google credentials and session cookies intercepted
If the victim selects the Google option, the phishing page opens a pop-up containing a Google login page provided through VoidProxy.
VoidProxy operates as an AiTM reverse proxy between the victim’s browser and Google’s real login service. It relays the login process while capturing credentials, multi-factor authentication responses, and session cookies.
Both Kali365 and VoidProxy can help attackers bypass multi-factor authentication by capturing an authenticated session. In the device-code attack, the workflow can bypass MFA altogether because the victim authorizes the request through Microsoft’s device-login process.
How to reduce the risk
Employees should avoid entering Microsoft device authorization codes after following links from unfamiliar websites. They should also be cautious about event invitations that require a corporate sign-in to claim free software, subscriptions, shares, or other benefits.
Organizations can reduce exposure by restricting unnecessary device-code authentication and deploying phishing-resistant MFA, such as FIDO2 security keys. These measures can help prevent credential theft and session hijacking even when attackers use convincing event pages and trusted services.
Source: www.bleepingcomputer.com