An invitation to a livestream from OpenAI or a private briefing from SpaceX can now be a trap for a work login. Push Security, a browser-security company, reported on Nov. 5 that fake event pages built around those two brands, and around Anthropic, are being used to steal Microsoft and Google account access. Each page promises a reward for signing in, such as three months of ChatGPT Pro or Grok Heavy.
Tech news in one quick read: Get the most interesting stories in startups, tech and programming, each with a short TLDR, in one daily email from TLDR.
We may earn a commission if you sign up.
Both sign-in buttons on the pages are built to get around multifactor authentication. One captures a Microsoft session through a device code, and the other relays Google’s real login so the attacker can copy the password, the second-factor response and the session cookie.
Luke Jennings, Push’s vice president of research and development, wrote that the company detected and blocked visits to the pages on customer browsers in late October. The write-up, published by Push, describes an OpenAI page copied from the company’s Academy events site that advertises a Nov. 4 livestream on ChatGPT for marketing teams, and an invite-only SpaceX “Starship Product Briefing” on AI and Grok dated Nov. 12. A third lure, “Anthropic at AWS re:Invent,” had already gone offline by the time Push looked. BleepingComputer covered the findings as well.
Kali365 and VoidProxy, the two kits behind the sign-in buttons
The “Continue with Microsoft” button starts a device-code attack. The page asks Kali365, a phishing-as-a-service platform, for a code and shows it in a “Confirm access” dialog. The visitor is told to enter it on Microsoft’s genuine device login page, and once the visitor approves, Microsoft hands tokens to the operator’s session. Microsoft described the same technique in its Storm-2372 research, which dates the activity to August 2024.
The “Continue with Google” button works differently. It opens a pop-up served from a VoidProxy reverse proxy, a second phishing-as-a-service platform, which passes Google’s real sign-in flow through the attacker’s server. Everything the victim types is copied on the way: the password, the MFA response and the session cookie. The Microsoft branch also ends on a VoidProxy page, so the operator gets a second capture of the password and cookie.
Push believes one operator is behind the pages and is most likely a paying customer of both kits, since both sell subscriptions to many unrelated campaigns. It found no link to the developers of either platform. Its confidence that the event pages connect to two older kits it tracks, ZoomMock and MSDRoot, is medium-high and rests mainly on one domain that hosted both.
The lures also recycle material. Push noticed a code comment reading “Same as Claude” and leftover Claude references on pages meant to represent other brands, which suggests the operator copies a working template and only partly rebrands it. The operator probably runs more lures than the three Push saw, and the reward on offer changes with the brand, from ChatGPT Pro to Grok Heavy to a “SpaceX share.”
Invitations built to slip past email scanners
The delivery chain leans on services that security tools tend to trust. Push saw links routed through Google short links, Zapier’s zpr.io redirector and an open redirect on x.com, with a Cloudflare challenge guarding the lure domain so automated scanners never reach the page. The main lure domain was three days old when Push examined it, and the operator rotates domains every few weeks, which makes blocklists a weak defense.
Earlier waves came through legitimate platforms. Cofense researchers flagged invitations from a spoofed Anthropic sender on Oct. 29, covered by CSO Online, that used GoTo Webinar branding and a misspelled sender name before sending recipients to a Zoom-style registration page. Analyst Frank Dickson told CSO that fear of missing out is the oldest lure in the book and that the employee should not be the last line of defense.
Push connects the Nov. 5 pages to earlier waves from the same operator. ZoomMock, one of its kits, served fake Zoom event pages from May to October 2026, and a newer kit, MSDRoot, began appearing on Oct. 22. ANY.RUN had reported in July that real Zoom event pages carried “partner summit” lures for Meta, OpenAI and Anthropic that branched into device-code and proxy phishing, so the brand-name bait has been in circulation for months.
Blocking device codes and verifying events
Because the Microsoft branch has the victim sign in on Microsoft’s real site, a passkey does not stop it. Push says only refusing device codes offered by a website, or blocking the device-code flow in Conditional Access after auditing existing use, closes that path. Microsoft’s own guidance says the same: block the flow wherever possible and allow it only through Conditional Access where it is required. A passkey does stop the VoidProxy branch.
For individual employees, Push’s advice is to open the company’s own site to check whether an event exists rather than clicking an invitation, and never to type a code that a web page supplies. The Gblock summary adds that anyone who already signed in should review unfamiliar sessions in Google and Microsoft account settings and ask IT to revoke tokens.
The target is the work account rather than a personal one. A stolen Microsoft or Google session gives an intruder mail, files and calendars, and in many companies the same identity unlocks internal tools. That explains why the pages ask visitors to sign in with a work address and why free subscriptions are the bait.
No victim count has been published. What remains unknown is how many of the invitations reached inboxes before the lure domain turned over, and whether the Nov. 12 SpaceX date will bring a fresh round of mail.
This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.
Tech news in one quick read: Subscribe to TLDR →
We may earn a commission if you sign up.
More from Morning Overview
- The NSA says three phone features should be off whenever you aren’t using them
- Hurricane Hunter radar shows four warning signs that a tropical cyclone is about to strengthen, a University of Miami study found
- Tropical Storm Rachel is dumping up to 12 inches on four Mexican states on its way to major hurricane strength
- The FTC says Lens.com doubled the price shoppers saw in Google ads
