Cyberpediaby Brain

The fake sign-in window inside a web page

Browser-in-the-Browser: a booby-trapped page draws a fake sign-in window. Real cases from 2022 and 2026 and the habits that keep you from being fooled.

On this page
  1. 1In 30 seconds
  2. 2What it is
  3. 3Straight answers
  4. 4Inside the attacker's head
  5. 5What you see on screen
  6. 6Under the hood
  7. 7Case files
  8. 8Sources
  9. 9Cite this article

In 30 seconds

Where it happens
Google TAG observed the technique against Polish users of webmail and social media[2]. Proofpoint describes it in 2026 in campaigns targeting Microsoft 365 accounts[3].
What the attacker wants
Your credentials[1] and, in the 2026 campaign described by Proofpoint, the session cookies that open the account[3].
The reflex
Do not sign in from a window that pops up from an unexpected page or message. Go through a bookmark you created yourself[4] and prefer passkeys, which Microsoft recommends among phishing-resistant authentication methods[5].
If it is already done
Change the password immediately on the service concerned and on every service where you used it[4].
Article map

What it is

Browser-in-the-Browser (BitB) is a form of phishing in which the booby-trapped page builds a fake sign-in window itself. The window imitates the one that appears when you choose to sign in to a service with the credentials of a trusted site such as Google, Apple or Twitter. If the person enters their credentials there, the attacker harvests them[1].

In a post dated July 19, 2022, Google TAG says it first observed and described the technique in March, then saw it again in attacks by the Ghostwriter group[2]. In 2026, Proofpoint reports it in a campaign against American experts in artificial intelligence policy[3]. This article shows what an employee sees, how the window is built, three dated cases and the habits to adopt.

Straight answers

What is a Browser-in-the-Browser attack?

The attacker creates a convincing phishing page, then adds a fake single sign-on (SSO) pop-up window to it[1]. To do so, they modify the code of the phishing site so that it embeds another, separate web page[1]. Proofpoint considers this form of phishing hard to detect[1].

Why does the window look genuine?

Microsoft describes the technique as simulating a legitimate browser window inside a web page[5]. In the July 2026 campaign, the overlay that is displayed is a fake Chrome browser window[3]. What looks like a browser window is in reality content of the page being visited[1].

Does multifactor authentication protect you?

Not in the July 2026 campaign described by Proofpoint. Behind the fake window, the page relays the sign-in to the genuine Microsoft infrastructure: the password, the multifactor authentication code and the conditional access checks all succeed, and the attacker captures the resulting session cookies[3]. MITRE ATT&CK notes that session cookies can bypass some multifactor authentication protocols, and cites Evilginx2 among the open-source tools that capture them through a malicious proxy[6].

How do you protect yourself?

Public bodies and vendors recommend three habits:

  • Reach the service through a bookmark you created yourself and, if in doubt, contact the organization directly where possible[4]. CISA gives the same advice: do not click the link in a message, look for the company's website or type the address yourself if you know it with certainty[7].
  • Check an unexpected request through another, independent channel. This is Proofpoint's instruction to people targeted by TA419, who should treat an unsolicited approach as a possible pretexting step[3].
  • Use passkeys. Proofpoint recommends that targeted organizations consider phishing-resistant authentication methods bound to the origin of the site, such as passkeys[3]. Microsoft cites FIDO keys and Microsoft Authenticator with a passkey[5].

Proofpoint also advises caution with shortened addresses, which can hide the true identity of a link[1].

What if I have entered my credentials?

Change the password immediately on the service concerned, and on every other service where you used the same one[4]. Keep the phishing message you received, as it serves as evidence[4]. If in the slightest doubt, contact directly, where possible, the organization whose identity was usurped[4].

Inside the attacker's head

The attacker is after access to the account. With a valid session cookie, they can sign in to a web service as an authenticated user, without credentials[6].

The TA419 group, which Proofpoint describes as China-aligned and driven by espionage, works in stages. It first sends a harmless message, for example an invitation to join an "AI Policy Advisory Committee", to get a reply. Once the target has replied, it sends a shortened address supposed to give further information[3]. The first domain controlled by the group acts as a filter: it runs a Cloudflare Turnstile check behind a fake OneDrive loading screen before redirecting to the phishing page[3].

In another campaign, Microsoft describes emails written with generative artificial intelligence and tailored to the victim's role: requests for proposals (RFPs), invoices, manufacturing workflows[5].

What you see on screen

In the July 2026 TA419 campaign, the person first goes through a fake OneDrive loading screen, then a OneDrive folder listing containing bait documents[3]. A click on a document, or OneDrive showing its own access-denied banner, brings up the fake Chrome window[3]. The sign-in that takes place there is relayed in real time to Microsoft[3].

In the campaign described by Microsoft, the final page is either a BitB window or a blurred preview of a document with a "Verify identity" button that leads to the page microsoft.com/devicelogin, with a code displayed next to it[5]. The themes observed are document access, electronic signature and voice messages[5].

Under the hoodIT teams

The July 2026 TA419 campaign relies on Frameless BitB, an open-source Browser-in-the-Browser kit. It contains the BitB overlay, an Evilginx template for Microsoft 365 and server-side substitution rules that inject the kit into the relayed pages[3]. The page received is the real authorization response from Microsoft, relayed in real time, and the proxy injects two malicious scripts into it[3].

  • The first script attaches a Shadow DOM to an element of the page and fills it with a OneDrive folder listing. The real documents are hosted in an OneDrive account belonging to the attacker[3].
  • The second script, written by the group, reports the victim's position in the sign-in to the attacker and gives a live view of each session. It automatically accepts "Keep me signed in" to extend the stolen session, and automatically submits one-time codes as soon as they are valid[3].

The group uses Cloudflare's content delivery network to hide the hosting IP address of its domains, which are often registered through NameSilo[3].

A study submitted to arXiv on May 25, 2025 describes a variant, QR-triggered BitB: a merger of BitB and QR code phishing, implemented with malicious prompts sent to the Google Gemini language model[8].

Case filesReal cases

July 19, 2022: Ghostwriter against Polish users

Google TAG says that Ghostwriter/UNC1151, an actor attributed to Belarus, remained active against webmail and social media accounts of Polish users. The group continues to use the Browser-in-the-Browser technique that TAG had first observed and described in March. A screenshot in the post shows an example aimed at Facebook users[2].

April 6, 2026: a device code phishing campaign

Microsoft describes a widespread campaign that hijacks device code authentication to compromise organizations' accounts at scale[5]. BitB is only one part of it: the final page may display the redirection to the Microsoft sign-in page as a BitB window, a technique Microsoft says the actor frequently uses[5]. The core mechanism is different. The victim enters a code and unknowingly authorizes the attacker's session, without their credentials being exposed[5]. A device code is valid for only 15 minutes; the attacker has it generated at the moment the link is clicked, so that it is always active[5]. A script often copies the code to the victim's clipboard[5]. After the compromise, Microsoft observes the creation of inbox rules to redirect or hide messages, and reconnaissance through Microsoft Graph[5]. Microsoft recommends blocking the device code flow wherever possible[5].

July 2026: TA419 against AI policy experts

Proofpoint reports that from July 8, 2026, TA419 impersonated the former principal deputy director of the White House Office of Science and Technology Policy, then an economist and foreign policy expert, to target AI policy experts at American think tanks, universities and law firms[3]. The invitations offered to join a fictitious committee or to contribute to a Senate Foreign Relations Committee report on AI export controls[3]. The chain ended on a fake OneDrive phishing page of the adversary-in-the-middle type, with the BitB kit described above[3]. In February 2026, the group had impersonated a senior Anthropic employee, for a similar phishing chain[3]. Proofpoint has observed this group since at least April 2025[3].

Sources

  1. Proofpoint, “How to Help Your Users Build Secure Browsing Habits”, 26 July 2023. https://www.proofpoint.com/us/blog/email-and-cloud-threats/help-users-build-strong-secure-browsing-habits (accessed 14 November 2026)
  2. Google, “Continued cyber activity in Eastern Europe observed by TAG”, 19 July 2022. https://blog.google/threat-analysis-group/continued-cyber-activity-in-eastern-europe-observed-by-tag/ (accessed 14 November 2026)
  3. Proofpoint, “Hallucinating Credibility: China-Aligned TA419 Impersonates its Way into US AI Policy Circles”, 1 October 2026. https://www.proofpoint.com/us/blog/threat-insight/hallucinating-credibility-china-aligned-ta419-impersonates-its-way-us-ai-policy (accessed 14 November 2026)
  4. Cybermalveillance.gouv.fr, “Que faire en cas de phishing ou hameçonnage ?”, 10 January 2020. https://www.cybermalveillance.gouv.fr/tous-nos-contenus/fiches-reflexes/hameconnage-phishing (accessed 14 November 2026)
  5. Microsoft, “Inside an AI‑enabled device code phishing campaign”, 6 April 2026. https://www.microsoft.com/en-us/security/blog/2026/04/06/ai-enabled-device-code-phishing-campaign-april-2026/ (accessed 14 November 2026)
  6. MITRE ATT&CK, “Steal Web Session Cookie”. https://attack.mitre.org/techniques/T1539/ (accessed 14 November 2026)
  7. CISA, “Recognize and Report Phishing”. https://www.cisa.gov/secure-our-world/recognize-and-report-phishing (accessed 14 November 2026)
  8. arXiv, “Exemplifying Emerging Phishing: QR-based Browser-in-The-Browser (BiTB) Attack”, 25 May 2025. https://arxiv.org/abs/2505.18944 (accessed 14 November 2026)