Threat reportVulnerabilityTL-2026-4993

Microsoft Edge Logic-Bug Full Exploit Chain Enables Code Execution Without Memory Corruption (Pwn2Own Berlin 2026, Orange Tsai; CVE-2026-45495)

highACTIVE

Microsoft Edge Logic-Bug Full Exploit Chain Enables Code (TL-2026-4993), also tracked as The Last Proof of Human Craft, is a high-severity software vulnerability scored CVSS 8.8, first published 2026-11-01. It has no confirmed attribution, affects Microsoft Microsoft Edge (Chromium-based), references 3 CVEs (CVE-2026-45492, CVE-2026-45494, CVE-2026-45495), maps to 12 MITRE ATT&CK techniques (T1005, T1059.007, T1185), and is covered by 9 detection rules and 5 indicators of compromise.

CVSS
8.8/10High
CVEs
3Referenced vulnerabilities
Techniques
12MITRE ATT&CK
Actors
0Not attributed
Detection rules
9SPL · KQL · Sigma
IOCs
5Indicators of compromise

Key facts for TL-2026-4993

Threat ID
TL-2026-4993
Also known as
The Last Proof of Human Craft, Edge Full Chain Pwn2Own Berlin 2026
Severity
HIGH
CVSS
8.8 (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H)
Status
ACTIVE
Category
VULNERABILITY
First published
Last reviewed
Attribution confidence
LOW
Motivation
UNKNOWN
Target sectors
technology, government administration, finance, health, critical-infrastructure, all-sectors-using-edge
Target regions
Global
Detection rules
9
Indicators of compromise
5

How Microsoft Edge Logic-Bug Full Exploit Chain Enables Code works

Orange Tsai (DEVCORE) publicly documented a full Microsoft Edge exploit chain built entirely from browser and OS logic flaws rather than memory corruption, achieving sandbox escape and code execution. It was demonstrated against fully patched Edge on Windows 11 25H2 at Pwn2Own Berlin 2026 and the write-up was published 2026-10-28. Secondary reporting associates the chain with CVE-2026-45495, fixed in Edge 148.0.3967.70.

Orange Tsai of DEVCORE Research Team demonstrated this chain against fully patched Microsoft Edge on Windows 11 at Pwn2Own Berlin 2026 (OffensiveCon, Berlin, May 14-16, 2026), earning $175,000 and 17.5 Master of Pwn points toward DEVCORE's first-place finish. On 2026-10-28 he published the full technical write-up, 'The Last Proof of Human Craft: Forging a Microsoft Edge Full Chain with Logic Bugs Only!', with accompanying slides and a video demonstration. The chain was developed entirely through manual research with no AI assistance and contains no memory-corruption primitive (no ROP, type confusion, or use-after-free) -- every step is a logic flaw in Edge's navigation, origin-checking, or privileged-API handling, or in how Windows resolves protocol handlers.

At a high level, the published chain links: (1) a popup-blocker allowlist entry that lets an attacker-controlled page open a window that is later redirected off the allowlisted origin; (2) a navigation throttle tied to Edge's Entra ID Conditional-Access guidance feature that validates only the last-committed URL and does not check the opener relationship, letting the throttle be reached from an unrelated origin; (3) an overly permissive CORS configuration on an embedded Microsoft asset host that discloses the signed-in user's email to any origin via a credentialed fetch, made usable because Windows 11 25H2's OneAuth/OOBE flow auto-signs Edge profiles into a Microsoft account; (4) a browser-process navigation API that, when invoked without an explicit source tab, defaults to whatever tab the user currently has focused, producing a universal cross-site-scripting (UXSS) primitive attackers use to run script in a victim-chosen context; (5) the read:// Immersive Reader WebUI scheme, which -- unlike other internal schemes -- is not excluded from javascript: URL execution, giving the UXSS a privileged execution context; and (6) the privileged chrome.edgeFeedbackPrivate.zipDiagnosticLogFiles() extension API reachable from that context, whose filename/content handling is vulnerable to path traversal and embedded-NUL extension-stripping, enabling an arbitrary file write outside the intended diagnostics directory. That write is used to overwrite Edge's AutoLaunchProtocolsComponent protocols.json policy file, which governs which origins may launch external protocol handlers; after an edge://restart reload, the attacker's origin is authorized to invoke a handler (telnet://) whose underlying executable is resolved via a Windows PATH search that includes a user-writable directory (WindowsApps), so a planted attacker binary is launched instead of the legitimate one -- yielding arbitrary code execution outside the browser sandbox.

No in-the-wild exploitation of this chain has been reported. Original hunt sourcing (GBHackers, 2026-10-29; Orange Tsai's blog, 2026-10-28) stated no CVE IDs, CVSS scores, or patch details. Subsequent secondary reporting (comss.ru, rapid7, NVD-adjacent trackers, HKCERT bulletin) associates three CVE numbers issued in Microsoft's May 2026 Patch Tuesday Edge/Chromium advisory with this research cycle -- CVE-2026-45492 (full-chain RCE, CVSS 5.4), CVE-2026-45494 (spoofing/auth bypass), and CVE-2026-45495 (elevation of privilege / sandbox escape, reported with conflicting CVSS values across sources: 8.8 per several trackers' summary pages and a 9.8/CVSS:3.1 AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H vector per Rapid7's detail page) -- with Edge Stable 148.0.3967.70 (Chromium 148.0.7778.168), released 2026-05-15, cited as the fix. These CVE-to-bug mappings come from secondary aggregator sites rather than from Orange Tsai's own write-up or a located Microsoft Security Update Guide page, so attribution of which specific CVE ID corresponds to which individual logic bug in the chain is not confirmed from primary sourcing and should be treated as tentative pending direct MSRC confirmation.

MITRE ATT&CK techniques used in TL-2026-4993

Collection

T1005 Data from Local System; T1185 Browser Session Hijacking

Execution

T1059.007 JavaScript; T1203 Exploitation for Client Execution; T1204.001 Malicious Link

Initial Access

T1190 Exploit Public-Facing Application

Credential Access

T1539 Steal Web Session Cookie

Persistence

T1546 Event Triggered Execution

stealth

T1574.007 Path Interception by PATH Environment Variable

Defense Evasion

T1574.007 Path Interception by PATH Environment Variable

Resource Development

T1584 Compromise Infrastructure

resource-development

T1586 Compromise Accounts

Reconnaissance

T1589.002 Email Addresses

Affected products and versions in Microsoft Edge Logic-Bug Full Exploit Chain Enables Code

  • Microsoft — Microsoft Edge (Chromium-based)
    Vulnerable versions: before 148.0.3967.70
    Fixed in: 148.0.3967.70 (Chromium 148.0.7778.168) and later
  • Microsoft — Windows 11
    Vulnerable versions: 25H2 (OOBE OneAuth auto-sign-in behavior at time of research)

Remediation for Microsoft Edge Logic-Bug Full Exploit Chain Enables Code

Patches

  • Microsoft Edge Stable 148.0.3967.70 (Chromium 148.0.7778.168), released 2026-05-15, cited by secondary sources as the fix for the associated May 2026 CVE batch.

Immediate actions

  • Update Microsoft Edge to version 148.0.3967.70 or later (Chromium 148.0.7778.168+) on all managed endpoints.
  • Verify enterprise Edge update-deferral/WSUS policies are not blocking the May 2026 Edge security build.
  • Audit AutoLaunchProtocolsComponent/protocols.json and other Edge component-policy files for unexpected modification dates or unauthorized entries.
  • Restrict write access to user-writable PATH-resolvable directories such as %LOCALAPPDATA%\Microsoft\WindowsApps via application control / AppLocker to prevent planted executables from being resolved by protocol handlers.

Workarounds

  • Disable or tightly scope OneAuth automatic sign-in / OOBE auto-provisioning of Microsoft accounts into Edge profiles where feasible in managed environments.
  • Restrict or monitor use of Edge's Immersive Reader (read://) feature via enterprise policy where not required for business use.

Longer-term hardening

  • Deploy application control (WDAC/AppLocker) to block execution of unsigned or unexpected binaries from user-writable directories, independent of Edge patch status.
  • Treat browser popup-allowlist and navigation-throttle logic as security-relevant attack surface in internal threat modeling, not just memory-safety code paths.
  • Monitor for anomalous telnet:// or other rarely-used protocol-handler invocations originating from the browser process.
  • Review CORS configuration on first-party asset/API hosts embedded in browser chrome (e.g. new-tab-page) for credential-exposing misconfigurations.

CVEs associated with Microsoft Edge Logic-Bug Full Exploit Chain Enables Code

CVE-2026-45492, CVE-2026-45494, CVE-2026-45495

Weaknesses (CWE) in Microsoft Edge Logic-Bug Full Exploit Chain Enables Code

CWE-22, CWE-346, CWE-863, CWE-200

Timeline of Microsoft Edge Logic-Bug Full Exploit Chain Enables Code

  • Orange Tsai (DEVCORE) demonstrates the Microsoft Edge full-chain exploit live on Day One of Pwn2Own Berlin 2026 (OffensiveCon), earning $175,000 and 17.5 Master of Pwn points against fully patched Windows 11.
  • Microsoft releases Edge Stable 148.0.3967.70 (Chromium 148.0.7778.168), later cited by secondary sources as the fix for the associated CVE batch.
  • Pwn2Own Berlin 2026 concludes; DEVCORE finishes first overall with $505,000 and 50.5 Master of Pwn points across SharePoint, Exchange, Edge and Windows 11 entries.
  • CVE-2026-45492, CVE-2026-45494 and CVE-2026-45495 published / indexed by vulnerability trackers (NVD, Rapid7, HKCERT bulletin) in connection with the May 2026 Edge security update.
  • Orange Tsai publishes the full technical write-up 'The Last Proof of Human Craft: Forging a Microsoft Edge Full Chain with Logic Bugs Only!' on his blog, with accompanying slides and video, releasing the complete chain publicly (under 100 lines of code).
  • Reporting notes the research also received the Pwnie Award 2026 for Epic Achievement.
  • GBHackers and other outlets publish summaries of the disclosed exploit chain, noting it achieves code execution without memory corruption.

Sources cited for Microsoft Edge Logic-Bug Full Exploit Chain Enables Code

Detection coverage for TL-2026-4993

As of 2026-11-01, Threadlinqs Intelligence publishes 9 detection rule(s) for TL-2026-4993 across Splunk SPL, Microsoft KQL and Sigma, covering 5 indicator(s) of compromise. The whole corpus is readable without an account; a free account unlocks full detection query text in Splunk SPL, Microsoft KQL and Sigma; paid tiers add raw indicator values, correlation and the MCP server. Threadlinqs MCP server · View plans.

9 detection rules (Splunk SPL, Microsoft KQL, Sigma) · Blue and above. Compare plans
5 indicators of compromise · Red and above. Compare plans

Further reading

Threadlinqs Intelligence — Real-Time Threat Detection Platform

[ 0 threats ] [ 0 det ] [ CRIT: 0 ] [ HIGH: 0 ]
// threat_feed
$ sort --newest
Showing all threats

Live intelligence console

Threat level
Fig. 01 · Threat weatherIndexing the archive…
1 square = 1 threat · click to open

Latest Threats