CVE-2026-55945Patch(microsoft / edge_chromium)

LOWCVSS 4.2 · MEDIUM

Exploit discussion active in current signal (1 latest mentions)

Immediate actions

  • Patch microsoft edge_chromium systems immediately
  • Hunt for exploitation attempts and persistence artifacts
  • Increase monitoring for publicly documented tradecraft

Recommended action window: High priority (within 72h)

NVD description

Concurrent execution using shared resource with improper synchronization ('race condition') in Microsoft Edge (Chromium-based) allows an authorized attacker to disclose information locally.

2.0/ 10 priority

Sources & remediation

Weakness type (CWE)
CWE-362

Priority

LOW

Exploitation

NONE

PoC

YES

Patch

AVAILABLE

Momentum

STABLE

Are you affected?

If you run products in this scope, you should treat this CVE as relevant to your environment.

  • edge_chromium

Threat summary

  • Public PoC is present in monitored signal
  • Patch or workaround signal is available
  • 9 mentions across 5 observed days
  • Momentum state: stable

What's happening

  • PoC mentioned or linked in 1 signal
  • Patch or workaround mentioned in 5 signals
  • Technical details provided in 6 signals
  • Disclosure: 2 classified signals
  • General: 2 classified signals
  • Peaked 3d ago at 4 mentions (2026-09-19); latest day: 1
  • 9 total mentions across 5 days

Affected systems

Vendors
Products
edge_chromium

Deep dive

Activity timeline9 mentions / 5d
01234Mentions · 2026-09-16: 1Mentions · 2026-09-19: 4Mentions · 2026-09-20: 1Mentions · 2026-09-21: 2Mentions · 2026-09-22: 1PoC Mentioned / Linked · 2026-09-21: 1Patch / Workaround · 2026-09-19: 4Patch / Workaround · 2026-09-21: 1Technical Details · 2026-09-19: 3Technical Details · 2026-09-21: 2Technical Details · 2026-09-22: 109-1609-1909-2009-2109-22
Signal classification4 categories
Patch
444.4%
Disclosure
222.2%
General
222.2%
PoC
111.1%
Referenced assets2 URLs
Classification over time
DateTotalLabels
2026-09-161
Disclosure1
2026-09-194
Patch4
2026-09-201
General1
2026-09-212
Disclosure1PoC1
2026-09-221
General1
Full discourse9 posts
  • Gal Weizman@WeizmanGal
    General

    במחקר #BragJack, אני חושב שהמתקפה על Edge הייתה הכי מעניינת, כשחלק ממנה הניב את CVE-2026-55945 כשMicrosoft השיקו את הסוכן המובנה שלהם בEdge, הם פיתחו דף השקה ייעודי דרכו היה אפשר ללמוד את הפיצ׳ר החדש בדפדפן לפני הדיווח שלי, זה היה ״https://www[.]microsoft[.]com/en-us/edge״ 🧵 https://t.co/zA8eTLylxX

    Post summary

    The tweet discloses the discovery of CVE‑2026-55945 in Edge's built‑in agent as part of a broader research effort, referencing the Edge website, but provides no further technical details, proof‑of‑concept, exploit code, or remediation information.

    3211964.7K
    2.4K followersView on X
  • ♫Why♥Not♪@Python_s_
    PoC

    🚨 AI SECURITY ALERT DISCLOSED BY: Forever Security PRODUCTS: Google Chrome / Gemini Live Microsoft Edge Perplexity Comet Opera Neon Claude in Chrome CVEs: CVE-2026-0628 — Chrome CVE-2026-55945 — Microsoft Edge IMPACT: A malicious browser extension can abuse trusted browser-agent communication paths and hijack AI-assisted actions without additional user clicks after the extension is installed. AFFECTED VERSIONS: Chrome before 143.0.7499.192 Microsoft Edge before 150.0.4078.48 Specific vulnerable version ranges for Comet, Opera Neon and Claude in Chrome were not published in the research. EXPLOITATION STATUS: Working PoC confirmed. No confirmed in-the-wild exploitation reported. ✅ URGENT ACTION: Update Chrome and Edge immediately. Update Comet, Opera Neon and Claude in Chrome to their latest releases. Audit extensions with broad host access, declarativeNetRequest or debugger permissions and enforce extension allowlisting where possible. SOURCE: Forever Security — BragJack Technical Research https://forever.security/blog/bragjack-attack-hijacks-every-browser-agent/ #CyberSecurity #InfoSec #AISecurity #AIAgents #BrowserSecurity #ThreatIntel #Vulnerability #SecOps #Chrome #MicrosoftEdge #Perplexity #OperaNeon #Claude

    Post summary

    The alert discloses two CVEs affecting AI-enabled browsers, confirms a working proof-of-concept, reports no active exploitation, and urges immediate patching to specific versions.

    10010200
    226 followersView on X
  • InfraSecurity Services@infrasecserv
    Patch

    BragJack: one unvetted browser extension hijacked the built-in AI agents in Chrome, Edge, Opera Neon, Comet and Claude in Chrome. No prompt injection needed, it crossed straight into the agent's privileged channel. CVE-2026-0628 and CVE-2026-55945 patched. #InfoSec #AISecurity

    Post summary

    The text announces that CVE-2026-0628 and CVE-2026-55945 have been patched, describing a browser extension that hijacked built-in AI agents across multiple browsers by crossing into a privileged channel without prompt injection.

    1001053
    12 followersView on X
  • CloudSecurityAlliance@cloudsa
    General

    CISO Daily Briefing: RatHat uses an AI copilot to drive Android banking trojans live; BragJack hijacks AI agents across 5 browsers via prompt injection (CVE-2026-0628, CVE-2026-55945); two Codex sandbox escapes hit the host OS. A hallucinated Pentagon AI report nearly triggered a military boarding op; 84% of AI governance failures trace to bypassed human checkpoints, not missing policy. https://labs.cloudsecurityalliance.org/research/ciso-daily-briefing-20260922/

    Post summary

    The text is a CISO daily briefing that mentions specific CVEs (CVE-2026-0628, CVE-2026-55945) in the context of prompt injection and sandbox escapes, but lacks details on patches, PoCs, or confirmed active exploitation.

    00010384
    18.9K followersView on X
  • THE TELL@TheTell_News
    Patch

    Google and Microsoft resolved the flaws they were assigned: CVE-2026-0628 and CVE-2026-55945. For Comet, Opera Neon and Claude in Chrome the report says only that similar flaws were shown. The practical bit: delete extensions you don't recognize or no longer use.

    Post summary

    The text confirms that Google and Microsoft have resolved the assigned vulnerabilities (CVE-2026-0628, CVE-2026-55945) and advises users to remove unrecognized browser extensions as a practical mitigation step.

    1000039
    11 followersView on X
  • Cyb3rVolt3x@AndraxPentester
    Patch

    The extension never has to be the agent. BragJack shows DNR can still reach the privileged component that reads files and takes actions, so Prompt Forcing just looks like the assistant working. Chrome and Edge patched their bugs (CVE-2026-0628, CVE-2026-55945). What stays is the install itself: “read and change all sites” is an agent control plane, not a content script.

    Post summary

    The text discusses privilege-escalation vulnerabilities (CVE-2026-0628, CVE-2026-55945) in Chrome and Edge browser extensions, confirming patches were applied while noting the underlying architectural issue with extension permissions remains.

    1000099
    45 followersView on X
  • ethereagle · building@ethereaglehq
    Disclosure

    BragJack: one ordinary extension commanded Claude in Chrome, Gemini Live, Edge, Comet, and Opera Neon. Forever Security took over the page the agent body already trusts, then sent it orders. Chrome got CVE-2026-0628. Edge got CVE-2026-55945. Comet, Neon, and Claude in Chrome have no CVE in their table. if you run those agents, check chrome://extensions for declarativeNetRequest you didn't mean to grant.

    Post summary

    The text discloses two CVEs (CVE-2026-0628 for Chrome, CVE-2026-55945 for Edge) linked to a browser extension-based attack chain, advises checking declarativeNetRequest permissions, but provides no PoC, exploit tool, patch, or active-exploitation claim.

    1000052
    307 followersView on X
  • General Intels Daily@intels_daily
    Disclosure

    🟠 𝗛𝗜𝗚𝗛 · 𝗗𝗶𝘀𝗰𝘂𝘀𝘀𝗶𝗼𝗻 🏢 Target: 𝗚𝗼𝗼𝗴𝗹𝗲, 𝗠𝗶𝗰𝗿𝗼𝘀𝗼𝗳𝘁, 𝗣𝗲𝗿𝗽𝗹𝗲𝘅𝗶𝘁𝘆 🧩 Products: 𝗚𝗼𝗼𝗴𝗹𝗲 𝗖𝗵𝗿𝗼𝗺𝗲, 𝗠𝗶𝗰𝗿𝗼𝘀𝗼𝗳𝘁 𝗘𝗱𝗴𝗲, 𝗢𝗽𝗲𝗿𝗮 𝗡𝗲𝗼𝗻 🛡️ CVE-2026-0628, CVE-2026-55945 A security researcher, Gal Weizman, disclosed a new attack technique called 'BragJack' that allows malicious browser extensions to hijack AI agents integrated into popular browsers (Chrome, Edge, Opera, Perplexity, Claude). The attack uses 'Prompt Forcing' to manipulate browser-level AI components, potentially allowing unauthorized access to local files, browsing history, and camera/microphone. Two CVEs (CVE-2026-0628, CVE-2026-55945) were identified and patched by vendors. #ThreatIntel #CTI

    Post summary

    The text announces a new attack technique called BragJack affecting browser AI agents, disclosing two CVEs and describing the vulnerability's impact without providing specific remediation details.

    00000203
    692 followersView on X
  • BrassBean@zher
    Patch

    Chrome's earlier related GlicJack path: fixed in Chrome ≥143.0.7499.192 (CVE-2026-0628, ~8.8). Edge: CVE-2026-55945 fixed ≥150.0.4078.48. other vendors paid bounties (~$20k across five, per the write-ups). research demo — not claiming a mass campaign in the wild. soft takeaway (no product pitch): 1. update the browser 2. audit extensions — especially anything with "read and change all your data on all websites" 3. ask: if this AI agent got forced… what can it actually reach? not ClickFix. not Hacktron SSO. not BlueMoon. not Shai-Hulud. different beat. you pick what stays installed.

    Post summary

    The post announces that Chrome and Edge have released fixes for CVE-2026-0628 and CVE-2026-55945 and urges users to update their browsers and audit extensions.

    0000039
    30 followersView on X
CPE platform detail1 entries

1 of 1 entries

PartVendorProductVersionTarget SWTarget HW
Appmicrosoftedge_chromium---

Explore more