Scammers are exploiting Google's legitimate security notification system to target Gmail users with fraudulent messages. Attackers inject malicious text into Google app password names, causing Google to generate authentic, digitally signed security alerts containing scam content—including fake invoices and fraudulent support numbers. This technique makes phishing emails appear to originate directly from Google's infrastructure, significantly increasing their credibility and success rate.
A patched ChatGPT vulnerability disclosed by Check Point Research could have allowed attackers to exfiltrate data from connected Gmail accounts and other integrated services using a covert cross-account channel via a shared internal JFrog Artifactory service. The flaw exploited weak isolation between containers supporting ChatGPT's code-execution features, enabling attackers to embed hidden instructions in shared conversations or custom GPT configurations. OpenAI decommissioned the vulnerable Artifactory instance; the cross-account channel is no longer active.
A large-scale phishing campaign is abusing Google's legitimate services including Meet, Search, DoubleClick, and Analytics as trusted redirect networks to bypass email security controls, deliver highly personalized credential-harvesting pages, and deploy ScreenConnect remote-access software. Attacks target organizations in manufacturing, government, finance, and non-profit sectors using DocuSign, Microsoft, FedEx, and QuickBooks impersonation. URL hash fragments encode victim email addresses to evade server-side logging and sandboxes.
Threat actors distributed a fake Google Gemini installer via Google Colab in July 2026 to deliver the Vidar information stealer. The campaign targeted users seeking AI tools, redirecting them through a spoofed Windows Software Hub to download malware that harvested browser credentials, including saved Gmail passwords and session data. Darktrace detected the compromise through behavioral analysis and blocked C2 communications.
Security researcher Gareth Heyes disclosed CSS bomb vulnerabilities in Gmail and other webmail services that exploit improper HTML/CSS sanitization. Attackers can craft malicious emails leveraging CSS selectors and mutations to exfiltrate authentication tokens, bypass image proxies to reveal user IP addresses, and spoof login interfaces to capture passwords. Multiple attack techniques work by manipulating the gap between sanitizer expectations and browser rendering of untrusted email content.
Security researchers demonstrated an indirect prompt-injection vulnerability in Claude's Chrome browser extension that enables attackers to intercept Gmail verification codes and authentication secrets. An attacker can craft malicious email content that, when summarized by the Claude agent, executes arbitrary JavaScript within the victim's authenticated Gmail session to query the Atom feed, extract verification codes, and hijack accounts on Slack, X, and Claude.ai. The attack exploits Claude's access to untrusted email content combined with powerful browser permissions.
Security researchers revealed attacks called "Pass-ta-key" that enable malware on compromised Windows devices to hijack Google accounts protected by synced passkeys, bypassing password and biometric authentication. Three techniques exploit Chrome's passkey storage, device trust mechanisms, and recovery workflows to gain valid authentication assertions. The threat requires malware already running on a victim's account but not administrator privileges.
GhostWriter, a state-sponsored threat actor, is conducting phishing attacks targeting personal Gmail accounts of Polish users. The campaign aims to compromise individual Gmail users, demonstrating espionage interest in this service. Attacks show signs of operational security issues, with some phishing emails misdirected to unrelated individuals with similar names.
A phishing campaign is targeting cryptocurrency users with emails that impersonate legitimate Google security alerts, leveraging Google account recovery wording such as "recovery contact request" and "review request" to create urgency. Reporting indicates the messages can appear to arrive through genuine Google systems rather than obviously suspicious senders, increasing their credibility. Attackers also use oversized blank space and hidden formatting in the email body to push malicious links below the initially visible content, helping them evade casual inspection. Linked landing pages are designed to harvest passwords, session cookies, and two-factor authentication codes, enabling rapid takeover of exchange accounts and wallets. Binance reported blocking 22.9 million phishing and scam attempts in Q1 2026, indicating the broader scale of activity into which this campaign fits. Gmail users, particularly those holding crypto assets, should treat unsolicited Google security or account-recovery emails with caution and verify any alerts directly in account settings.
Security researchers at LayerX disclosed a vulnerability in Anthropic's "Claude in Chrome" browser extension that lets any other installed Chrome extension — even one with no declared permissions — hijack the Claude AI assistant and use its authenticated session to read, forward, and delete the victim's Gmail messages, as well as access Google Drive and GitHub data. The root cause is a trust boundary flaw in the extension's externally_connectable handler, which validates only the origin (claude.ai) and not the script's execution context, allowing injected scripts to issue privileged commands. Researchers bypassed Claude's user-confirmation guardrails via approval looping and DOM-based perception manipulation, requiring no user interaction. LayerX reported the issue to Anthropic on April 27, 2026; Anthropic shipped version 1.0.70 on May 6, 2026 adding explicit approval flows, but researchers say the patch is incomplete and the flaw remains exploitable when the extension runs in privileged "Act without asking" mode or via the side-panel initialization flow. Gmail users who have installed the Claude in Chrome extension should update to the latest version, avoid privileged/auto-act mode, and audit other installed Chrome extensions.