Fake AI Event Invites Steal Microsoft and Google Logins
Push Security says fake OpenAI, SpaceX and Anthropic event invites steal corporate Microsoft and Google credentials via phishing kits that bypass multifactor authentication.
By Nathan Brooks
3 min read
What's News
- Push Security reported the campaign in late October 2026 after blocking malicious pages on customer browsers.
- Fake events included a November 4 OpenAI livestream offering three months of ChatGPT Pro and a November 12 SpaceX briefing.
- The Microsoft path uses Kali365 device codes plus VoidProxy to steal tokens, passwords and session cookies without needing the password.
- Two phishing kits, ZoomMock and MSDRoot, appear to share operators, though the link is not conclusive.
- Microsoft recommends restricting device code authentication via Entra Conditional Access and adopting FIDO2 keys or passkeys.
A phishing campaign impersonating OpenAI, SpaceX and Anthropic is using fake AI event invitations to steal corporate Microsoft and Google account credentials, Push Security reported in late October 2026. The security firm said it detected and blocked its customers' browsers from reaching the malicious registration pages during that period.
The lure is simple and effective: free access to premium AI tools. Push Security documented three fraudulent events. A fake OpenAI event advertised a November 4 livestream titled "ChatGPT for Social & Digital Marketing Teams," offering participants three months of ChatGPT Pro. A supposed SpaceX invitation-only briefing on November 12 promised a choice between three months of Grok Heavy or a "SpaceX share." A third page, impersonating Anthropic at "AWS re 2026," was already inactive when researchers examined it.
Push Security stressed that OpenAI, SpaceX and Anthropic had nothing to do with these events. Each registration page asked visitors to sign in with a corporate Microsoft or Google account to "complete registration and claim the promised rewards" — and that sign-in is where the theft begins.
How the attack chain works
The campaign routes victims through legitimate services to hide the destination. Push Security found shortened links tied to Google and Zapier; earlier related attacks used Zoom event pages and GoTo Webinar email infrastructure. Once a victim lands on the malicious site, a Cloudflare security challenge appears — real technology that also slows automated security scanners.
Behind the challenge sit two buttons, "Continue with Microsoft" and "Continue with Google," and each triggers a different attack chain.
The Microsoft path connects to Kali365, a phishing-as-a-service platform. Kali365 generates a Microsoft device authentication code and tells the victim to copy it into Microsoft's genuine sign-in page. Because the attacker initiated the request, approving the code hands over authentication tokens that can unlock Microsoft 365 resources without the account password. Microsoft users then get redirected to a second phishing service, VoidProxy, which tries to grab the password and the authenticated session cookie.
The Google path opens a pop-up that looks like Google's sign-in screen but routes the login through VoidProxy as an intermediary. The service can capture credentials, intercept certain multifactor authentication responses, and steal session cookies — the digital passes that keep you logged in without re-entering a password.
Push Security identified two phishing kits behind the fake event pages, dubbed ZoomMock and MSDRoot. Evidence suggests the same operators built both kits, though the link isn't conclusive. Researchers found no connection between the developers of Kali365 and VoidProxy and believe the campaign operators are simply customers of both services. The fraudulent pages also auto-translate into a visitor's language, so the crew can target regions without building new sites.
What does this change for IT teams?
Plenty. The campaign combines trusted infrastructure — Google URL shorteners, Zapier, Cloudflare, Microsoft's own device authentication flow — with phishing-as-a-service, producing attacks that can bypass MFA. The Microsoft path is the nastiest part: the URL is genuine, only the request came from an attacker. Employees used to legitimate device sign-in flows have no visual red flag.
Some questions remain open. Push Security did not establish how many accounts, if any, were compromised, and researchers haven't confirmed the initial delivery method for every invitation in the newer campaign.
A Microsoft spokesperson said: "We take reports of malicious online activity seriously and continuously work to help protect our customers from evolving threats. Microsoft Defender includes detections to block the threats identified in the report." Google, OpenAI, Anthropic and SpaceX did not respond to requests for comment before the reporting deadline.
Microsoft recommends restricting device code authentication wherever it isn't necessary; admins can enforce that through Microsoft Entra Conditional Access policies after reviewing business needs. Push-oriented defenses go further: phishing-resistant authentication such as FIDO2 security keys or passkeys — though, as the reporting notes, no single measure stops every attack vector.
Source: epyhia.com
More from Nathan Brooks
Show full bio
News editor covering regulation, policy and their impact on entrepreneurship, startups and venture funding at Business Bearings.
2,230 articles