PortSwigger Researcher Reveals CSS Attacks That Break Webmail Defenses to Steal Passwords, Tokens

AI digestMimile.ai EditorialAugust 8, 2026 at 01:03 PM

This digest was compiled by AI from multiple sources — links to the originals are below.

PortSwigger Researcher Reveals CSS Attacks That Break Webmail Defenses to Steal Passwords, Tokens

PortSwigger researcher Gareth Heyes presented novel CSS attacks at Black Hat USA 2026 that can escape email boundaries to steal passwords and tokens from Outlook, Gmail, Yahoo Mail, AOL, Fastmail, and Proton Mail. Public proof-of-concept code remains online as of August 8, while some of the chained exploits still work. Fastmail and Proton Mail have fixed their vulnerabilities, but Outlook and Gmail remain vulnerable.

CSS Escape Methods

The research outlines two attack paths: leveraging HTML and CSS that webmail already permits, and exploiting discrepancies between sanitizer approvals and browser behavior. In one Outlook chain, a label element triggers controls outside the message, while application JavaScript converts sanitized attributes into new DOM nodes, allowing arbitrary CSS through a media-query parsing trick. This creates a fake Microsoft sign-in screen that captures passwords in real time, with Firefox resetting a one-second timer to maintain the deception. Another attack uses a paste race in Yahoo Mail and AOL Mail to briefly retain active CSS, enabling theft of a 12-character Medium login token that the attacker reconstructs from request data.

Vulnerability Status

Fastmail fixed two CSS mutation bugs, and a Proton Mail proxy bypass ceased working on retest. However, as of the August 6 publication date, Outlook label-jacking and Gmail's image-set() bypass still functioned. It remains unclear whether the full Outlook password-capture chain has been patched. The researcher recommends isolating HTML email in sandboxed iframes and restricting CSS, custom attributes, select menus, and image requests.

What's Next

Public proof-of-concept code is still available on August 8, and no reports of malicious exploitation have emerged. It is unclear whether affected providers will deploy the recommended mitigations before attackers adopt the techniques.

1 source

PortSwigger Researcher Reveals CSS Attacks That Break Webmail Defenses to Steal Passwords, Tokens

thehackernews.com

The Hacker News

Technology

iAuthFlow v2 malware adds attacker passkeys to hijacked email accounts

Security researchers at Abnormal discovered iAuthFlow v2, a malware toolkit sold on Russian forums for over $10,000 that lets attackers regain access to compromised email accounts even after password resets. The tool phishes credentials from Google, Microsoft, iCloud, or LinkedIn, then silently creates an attacker-controlled passkey. Abnormal advises users to audit passkeys, OAuth tokens, mail rules, and remove rogue authentication methods.

Technology

Chrome Passkey Theft Possible Without User Prompt, Research Shows

Palo Alto Networks Unit 42 released a technical analysis on August 3 detailing three methods that let malware on a compromised Windows machine silently steal and reuse passkeys from Chrome’s Google Password Manager. The Pass-ta-key, Silver Pass-ta-key, and Golden Pass-ta-key techniques exploit gaps in how Chrome stores device keys and handles re-enrollment, enabling attackers to obtain valid authentication assertions without triggering a fingerprint or PIN prompt. The research noted that none of the attacks break passkey cryptography, highlighting implementation flaws rather than protocol weaknesses.

Technology

Check Point uncovers OAuth phishing campaign exploiting genuine Microsoft logins, targeting 120+ orgs

Check Point's email research team disclosed a phishing campaign that ran from late June into July 2026, abusing real Microsoft login pages to compromise accounts across over 120 organizations worldwide. The attackers sent fake Microsoft Teams notifications directing victims to a legitimate OAuth authorization URL, then tricked them into approving permissions for a malicious application, granting access to email, files, and messaging platforms without stealing passwords or defeating multi-factor authentication. The tactic eliminates the need for look-alike domains or credential harvesting, underscoring a shift in attacker methods.

Technology

Researchers say AI agents installed malicious code via hallucinated docs

Researchers found that AI agents including Claude, OpenAI Codex, and Nous Research's Hermes installed suspicious code from unregistered packages referenced in website documentation. The team registered some unclaimed names and observed a Fortune 500 company ping within an hour. The findings highlight risks when AI agents are granted permission to execute commands.

PortSwigger Researcher Reveals CSS Attacks That Break Webmail Defenses to Steal Passwords, Tokens

CSS Escape Methods

Vulnerability Status

What's Next

Read more

Read more