Cybercriminals have figured out how to hijack Microsoft 365 accounts without ever touching a password—even if you've completed multi-factor authentication. A technique called ConsentFix, detailed by Kaspersky in July 2026, combines a legitimate Microsoft login with a simple drag-and-drop instruction that hands an attacker a valid authorization token, granting access to your email, files, and more.

How ConsentFix Tricks You Into Authorizing the Attacker

The attack begins with a familiar lure. You receive an email that appears to come from a colleague or a service like SharePoint, asking you to review a document. The link takes you to a page that mimics a file-sharing service. To view the file, you're prompted to “Verify with Microsoft.”

This is where most people’s security instincts fail. Clicking the button sends you to a real Microsoft sign‑in page. You enter your work credentials and approve the MFA prompt—maybe via authenticator app, SMS, or a passkey. Everything looks legitimate because it is: the domain, the certificate, and the authentication flow are all genuinely Microsoft. You’ve done nothing wrong.

But after you sign in, nothing happens. No document opens. Instead, the browser shows an error or a blank page. The address bar now displays a localhost URL, typically containing a long string of characters—an OAuth authorization code. The phishing page, which you still have open, instructs you to return and drag the information icon (ⓘ) from the address bar into a designated drop zone on the page.

That single drag copies the entire localhost URL onto the attacker’s infrastructure. Instantly, their server extracts the authorization code and exchanges it for access and refresh tokens. From that moment, the attacker can access your Microsoft 365 resources—email, Teams, OneDrive, SharePoint—without your password or any further MFA prompt.

What Attackers Can Do with a Stolen Token

The blast radius depends on your organization’s Microsoft 365 license and your own privileges. At minimum, the attacker gains full access to your mailbox. Years of archived messages, attachments, and contact lists become an open book. They can search for financial details, legal documents, or password reset emails, then use that intelligence to craft convincing impersonation attacks.

Beyond email, attackers typically gain entry to Microsoft Teams. They can read your direct messages, join meetings as you, and send malicious links to your colleagues—all from your trusted account. OneDrive and SharePoint expose every file you have access to, allowing exfiltration of confidential data or even planting of malware for others to download.

For organizations running Azure or connected SaaS apps, the compromise can spread further. If your account holds administrative roles, the attacker can modify security settings, create new accounts, or move laterally across cloud resources. Privileged users—finance, IT, executives—should assume a token theft gives an intruder the keys to far more than just their own inbox.

Why MFA Couldn’t Save You

Multifactor authentication protects the login ceremony. It confirms that the person typing the password is the real account holder. But ConsentFix doesn’t steal your password; it hijacks the session after you’ve proven your identity. Microsoft’s servers see a legitimate user authenticating through a legitimate application. The problem is that you then hand the resulting authorization to someone else.

This gap between authentication and authorization is critical. ConsentFix exploits the trust you place in the entire Microsoft ecosystem: the real login page, the MFA prompt you’ve been trained to approve, and even legitimate first‑party Microsoft applications that attackers sometimes use as the OAuth client. When you see “Microsoft wants to access your data,” it may truly be Microsoft—but the person on the other end of that transaction isn’t you.

How We Got Here: The ClickFix Evolution

ConsentFix didn’t appear in a vacuum. It builds on the ClickFix social‑engineering wave that surged in 2024. Early ClickFix campaigns showed victims fake browser errors or CAPTCHAs, then told them to press Windows+R, paste a command into the Run dialog, and execute it. That executed malicious PowerShell or mshta scripts, leading to infostealers like Lumma or Vidar.

Those attacks required compromising the endpoint—installing malware, modifying the registry, or establishing persistence. ConsentFix streamlines the process: no executable lands on the PC, and the victim never pastes a command. Instead, the attacker tricks the user into completing a legitimate OAuth flow and revealing the authorization code. Push Security researchers first documented the technique in late 2025, and by May 2026 criminal forums were sharing a more automated version dubbed ConsentFix v3.

The shift matters because it turns identity itself into the primary attack surface. Defenders can no longer rely solely on endpoint detection or password hygiene. Once a token is extracted, the attacker operates from Microsoft’s own cloud, often bypassing legacy network and device controls.

What to Do Now: Layered Defenses for Users and Admins

ConsentFix can’t be stopped by disabling a keyboard shortcut or blocking PowerShell. It preys on human behavior, and defense requires a combination of user awareness, cloud policy, and rapid response.

If you’re a user:
- Never drag information from the address bar, copy a localhost URL, or paste anything into a command prompt because a website tells you to “fix” something. Legitimate Microsoft services will never ask you to do this.
- If a document fails to open after a Microsoft login, close the browser tab and contact your IT department. Do not return to the requesting page.
- Pay attention to odd browser behavior after sign‑in. An immediate redirect to a localhost address is a red flag.

If you’re an administrator:
- Immediately revoke sessions and refresh tokens for any account suspected of compromise. A password reset alone is not enough—existing tokens will still work.
- Audit OAuth grants and enterprise applications in Microsoft Entra ID. Look for suspicious permission assignments, newly created service principals, or unusual consent records.
- Enforce Conditional Access policies that require a compliant, managed device for all cloud resources. This makes stolen tokens much harder to use from an unmanaged attacker machine.
- Deploy token protection if your license supports it. This binds access tokens to the device they were issued to, preventing replay on other hardware.
- Enable detailed sign‑in and audit logging in Microsoft Defender XDR or Sentinel. Watch for an unusual client application requesting tokens, rapid token use from a different geography, or mass Graph API calls right after a legitimate sign‑in.

For long‑term hardening:
- Limit which applications users can consent to. Move to an admin consent workflow so IT can vet every third‑party app that requests permissions.
- Train staff on the specific mechanics of ConsentFix—not just “don’t click suspicious links.” Show them the drag‑and‑drop lure and explain why a real authentication prompt will never include that step.
- Separate administrative accounts from everyday productivity. A privileged user should never browse email or open documents in the same browser session they use for Azure management.

What to Watch For

Attackers are already refining ConsentFix. Newer variants may replace the drag gesture with a QR‑code scan or a device‑code flow that feels even more routine. Automation will reduce the steps the victim must perform, making the attack harder to notice.

Browser vendors may start flagging drag‑and‑drop operations that expose sensitive address‑bar content, especially when the target is a localhost URL containing an authorization code. Microsoft can also tighten its OAuth client policies and detect improbable token redemptions more aggressively.

But the real shift must happen in how we think about identity security. ConsentFix proves that no matter how strong your MFA is, the moment you authorize the wrong transaction, the game is over. Defenders must treat authentication intent as a first‑class security control—because attackers certainly treat it as an opportunity.