- Hackers used fake OpenAI, SpaceX and Anthropic event invitations to lure employees into phishing attacks.
- Push Security blocked visits to the fake registration pages on customer browsers in late October 2026.
- The pages used Kali365 for Microsoft device-code phishing and VoidProxy for Google reverse-proxy phishing.
The pages asked visitors to sign in with work accounts to claim rewards including three months of ChatGPT Pro, three months of Grok Heavy or a SpaceX share. The campaign used email-delivered invitations that led to adversary-in-the-middle and device-code phishing, with flows targeting Microsoft and Google accounts.
Push Security found pages advertising an OpenAI Academy livestream, a private SpaceX briefing and an Anthropic event at AWS re:Invent. It said the event-themed pages were built by the campaign operator, separately from the Kali365 and VoidProxy phishing-as-a-service platforms used in the login flows.
On this page
Kali365 and VoidProxy captured sign-in data through separate flows
Push Security said the Microsoft button started a Microsoft device-code login through Kali365. After a victim entered the code, signed in and approved access, Microsoft issued tokens to the operator’s Kali365 session.
The Microsoft flow then sent the victim to a VoidProxy Microsoft sign-in page that captured the password and session cookie. The Google button opened a VoidProxy reverse proxy of Google’s sign-in page, which captured the password, the MFA response and the session cookie.
Push described Kali365 and VoidProxy as separate phishing-as-a-service platforms designed to bypass or defeat MFA checks. It said it had not seen anything linking their developers.
Fake pages promoted rewards and named events
The OpenAI page advertised a Nov 4 livestream titled “ChatGPT for Social & Digital Marketing Teams.” The SpaceX page advertised an invite-only “Starship Product Briefing” on Nov 12 about SpaceX AI for marketing and Grok.
Push linked ZoomMock and MSDRoot to the event-page campaign
Push said MSDRoot pages started on Oct 22 and identified ZoomMock and MSDRoot as the kits used to build and deploy the fake event-invitation lure pages. It rated the link between the two kits as medium-high.
Push said the operator changed themes from Zoom to Anthropic, OpenAI, SpaceX and Claude between Oct 20 and Oct 30. The company said the event-themed pages were separate from the Kali365 and VoidProxy platforms.
Push Security said its behavioral phishing detection identifies the Kali365 device-code page and VoidProxy reverse proxy when they render, using real-time analysis. It said customers did not need to take further action.
Push tracked the lure-page kits as ZoomMock and MSDRoot, and said the operator moved from Zoom to Anthropic, OpenAI, SpaceX and Claude between Oct 20 and Oct 30. It rated the link between ZoomMock and MSDRoot as medium-high.
What to do now
- Staff should verify invitations by visiting an organiser’s official site directly instead of following an unexpected registration link.
- Security teams should review sign-in logs for unusual device-code approvals, unfamiliar sessions and unexpected OAuth activity.
FAQ
What happened in the fake OpenAI, SpaceX and Anthropic event campaign?
Hackers used fake event invitations and registration pages to lure employees into phishing attacks. Push Security detected and blocked visits to the pages on customer browsers in late October 2026.
Which accounts did the phishing pages target?
The pages used Kali365 to phish Microsoft accounts and VoidProxy to phish Google accounts.
What information could the phishing flows capture?
In the Microsoft flow, Microsoft issued tokens to the operator’s Kali365 session after the victim approved, and a VoidProxy page captured a password and session cookie. The Google flow captured the password, MFA response and session cookie.
What are Kali365 and VoidProxy?
They are separate phishing-as-a-service platforms designed to bypass or defeat MFA checks. Push said it had not seen anything linking their developers.
What did TechBooky recommend for unexpected event invitations?
TechBooky recommended verifying invitations by visiting an organiser’s official site directly instead of following an unexpected registration link. It also recommended reviewing sign-in logs for unusual device-code approvals, unfamiliar sessions and unexpected OAuth activity.
Sources
- Fake OpenAI, SpaceX events used to steal Microsoft, Google accounts, HKCERTPrimary
- Fake event invites combine Kali365 and VoidProxy payloads, Push SecurityPrimary
- Fake OpenAI Codex download tricks macOS users into installing malware, Help Net Security
- Fake OpenAI and SpaceX Event Invites Target Work Accounts, TechBooky
How we checked this story
| Claim | Source | Status |
|---|---|---|
| HKCERT said hackers were using fake OpenAI, SpaceX, and Anthropic event invitations to lure employees into phishing attacks. | HKCERT | Confirmed |
| Push researchers found fake registration pages for OpenAI, SpaceX, and Anthropic events. | Push Security | Confirmed |
| Push said the pages used Kali365 to phish Microsoft accounts and VoidProxy to phish Google accounts. | Push Security | Confirmed |
| Push detected and blocked visits to the fake event registration pages on customer browsers in late October 2026. | Push Security | Confirmed |
| The pages advertised an OpenAI Academy livestream, a private SpaceX briefing, and an Anthropic event at AWS re:Invent. | Push Security | Confirmed |
| The pages offered three months of ChatGPT Pro, three months of Grok Heavy, or a SpaceX share as rewards. | Push Security | Confirmed |
| The campaign used email-delivered invitations leading to AiTM and device-code phishing. | Push Security | Confirmed |
| The Microsoft button started a Microsoft device-code login through Kali365. | Push Security | Confirmed |
| The Google button opened a VoidProxy reverse proxy of Google’s sign-in page. | Push Security | Confirmed |
| Kali365 and VoidProxy were separate phishing-as-a-service platforms designed to bypass or defeat MFA checks. | Push Security | Confirmed |
| The event-themed pages were built by the campaign operator, separately from Kali365 and VoidProxy. | Push Security | Confirmed |
| Push tracked the lure-page kits as ZoomMock and MSDRoot, with MSDRoot pages starting on October 22. | Push Security | Confirmed |
| The OpenAI page advertised a livestream on November 4 titled ChatGPT for Social & Digital Marketing Teams. | Push Security | Confirmed |
| The SpaceX page advertised an invite-only briefing on November 12 about SpaceX AI for marketing and Grok. | Push Security | Confirmed |
| The Microsoft flow gave tokens to the operator’s Kali365 session after the victim entered a device code, signed in, and approved access. | Push Security | Confirmed |
| The Microsoft flow then sent victims to a VoidProxy page that captured the password and session cookie. | Push Security | Confirmed |
| The Google proxy captured the password, MFA response, and session cookie. | Push Security | Confirmed |
| Push rated the link between ZoomMock and MSDRoot as medium-high. | Push Security | Confirmed |
| Push said it had not seen anything linking the Kali365 and VoidProxy developers. | Push Security | Confirmed |
| Push said the operator moved from Zoom to Anthropic, OpenAI, SpaceX, and Claude between October 20 and October 30. | Push Security | Confirmed |
| Push said its behavioral phishing detection identifies and blocks the Kali365 device-code page and VoidProxy reverse proxy in real time. | Push Security | Confirmed |
| TechBooky said Push detected and blocked visits to the pages in late October. | Push Security, via TechBooky | Attributed |
| Cato Networks said it observed active payload delivery for macOS but not Linux in the separate Codex campaign. | Cato Networks, via Help Net Security | Attributed |
| Help Net Security said Cato found strong similarities between the separate fake Codex delivery chain and Atomic macOS Stealer. | Cato Networks, via Help Net Security | Attributed |
| HKCERT published its report on Nov 5, 2026. | HKCERT | Confirmed |
| pushsecurity.com published its report on Nov 5, 2026. | pushsecurity.com | Confirmed |
Could not verify
- Whether any accounts were compromised is not established
- Whether OpenAI, SpaceX, or Anthropic systems were compromised is not established
- How many people received the invitations or visited the pages is not established
- Whether the campaign operator was identified is not established
- Whether the separate fake Codex campaign is linked to this event is not established
Related stories
AI-Powered AttacksJustice Department and partners disrupt Gootloader infrastructure
Vulns & PatchesCitrix discloses critical NetScaler zero-days after exploitation reports
Data LeaksArara email service breach exposed more than 1 million Osaka addresses