Skip to navigation

Threat reportVulnerabilityTL-2026-0870

Spyder (SiderAI) & MaXSS (MaxAI) Chrome/Edge Extension Message-Validation Flaws Enable Zero-Interaction Browser Session Compromise Across 11M+ Installs

criticalACTIVE

Rebora Security disclosed two critical improper-input-validation flaws in widely deployed AI browser extensions: 'Spyder' in SiderAI (10M+ installs, Chrome Web Store top-25) and 'MaXSS' in MaxAI (1M+ installs). A malicious webpage can drive each extension's content script to relay attacker-controlled messages to its privileged background service worker without origin/sender validation, granting the page the extension's full capabilities with no user interaction beyond visiting the page.

Key facts

IdentifiersTL-2026-0870
Also known asSpyder, MaXSS, MaXSS & Spyder
SeverityCRITICAL
Affected productsSider.AI Sider: ChatGPT Sidebar + AI Tools (SiderAI Chrome/Edge extension); MaxAI.me MaxAI.me (MaxAI Chrome/Edge extension)
Weaknesses (CWE)CWE-20, CWE-346, CWE-940, CWE-863
MITRE ATT&CK techniquesT1005 Data from Local System, T1059.007 JavaScript, T1068 Exploitation for Privilege Escalation, T1071.001 Web Protocols, T1113 Screen Capture, T1114.002 Remote Email Collection, T1176 Software Extensions, T1185 Browser Session Hijacking, T1189 Drive-by Compromise, T1203 Exploitation for Client Execution, T1217 Browser Information Discovery, T1528 Steal Application Access Token and 3 more in the attack flow
MotivationUNKNOWN
Target sectorstechnology, all sectors, consumer, enterprise, government
Target regionsGlobal
Detection rules9 (1 shown free)
Sources6 cited sources

Detection coverage

As of 2026-06-19, Threadlinqs Intelligence publishes 9 detection rule(s) for TL-2026-0870 across Splunk SPL, Microsoft KQL and Sigma, covering 16 indicator(s) of compromise. The whole corpus is readable without an account. Every threat comes with one complete detection in Splunk SPL, Microsoft KQL and Sigma, free for everyone; a free account adds daily TL Creds to unlock the rest. Paid plans add a larger daily allowance, correlation and the MCP server. Threadlinqs MCP server · View plans.

Free detection: Spyder/MaXSS — Content Script Forwards Unvalidated Page-Origin Message to Background (Exploitation)

Detects the core exploit primitive of the Spyder (SiderAI) and MaXSS (MaxAI) flaws: a Manifest V3 content script accepts a window.postMessage / page-injected DOM message from an untrusted web origin and forwards it to the privileged background service worker via chrome.runtime.sendMessage with no sender/origin validation (CWE-20/CWE-346). The query keys on browser extension messaging telemetry where a content_script forward event is marked validated=false (or where the message origin is an arbitrary web page rather than the extension's own context), scoped to the affected SiderAI/MaxAI extension IDs.

  • Languages: Splunk SPL, Microsoft KQL, Sigma
  • ATT&CK: T1189, T1059.007, T1176
  • Data sources: chrome:extension:messaging, browser_extension_telemetry, edr_browser_module
  • Severity: critical
  • Confidence: high

Splunk SPL

index=browser sourcetype="chrome:extension:messaging" stage=content_script event=forward_to_background
| eval ext_id=coalesce(extension_id,ext)
| where validated="false" OR (isnotnull(origin) AND NOT match(origin, "^chrome-extension://"))
| eval suspect_extension=if(ext_id IN ("difoiogjjojoaoomphldepapgpbgkhkb","mhnlakgilnojmhinhkckjpncpbhabphi"),"KNOWN_VULN","other")
| stats count min(_time) as first_seen max(_time) as last_seen values(origin) as page_origins values(data) as forwarded_cmds by host, user, ext_id, suspect_extension
| where count > 0
| sort - count

Microsoft KQL

BrowserExtensionMessaging
| where TimeGenerated > ago(24h)
| where Stage == "content_script" and Event == "forward_to_background"
| where Validated == false or (isnotempty(Origin) and Origin !startswith "chrome-extension://")
| extend SuspectExtension = iff(ExtensionId in ("difoiogjjojoaoomphldepapgpbgkhkb","mhnlakgilnojmhinhkckjpncpbhabphi"), "KNOWN_VULN", "other")
| project TimeGenerated, DeviceName, AccountName, ExtensionId, SuspectExtension, Origin, ForwardedData
| sort by TimeGenerated desc

Sigma

title: Browser Extension Content-Script Forwards Unvalidated Page-Origin Message to Background
id: 6f3b9a52-0a1d-4c8e-9f21-7c5d0e3a1b01
status: experimental
description: Detects a Manifest V3 content script relaying an attacker-controlled page-origin message to the privileged background service worker without sender/origin validation (Spyder/MaXSS pattern, CWE-20/CWE-346).
references:
    - https://rebora.io/blog/spyder-and-maxss-chrome-extension-vulnerabilities-put-millions-at-risk/
    - https://attack.mitre.org/techniques/T1176/
author: Threadlinqs Team
date: 2026/06/19
tags:
    - attack.initial_access
    - attack.t1189
    - attack.execution
    - attack.t1059.007
    - attack.persistence
    - attack.t1176
logsource:
    product: browser
    category: extension_messaging
detection:
    selection:
        stage: 'content_script'
        event: 'forward_to_background'
        validated: false
    condition: selection
falsepositives:
    - Extensions that legitimately bridge same-origin in-page widgets and tag messages validated=false due to incomplete instrumentation
level: high

Known false positives:

  • Extensions that legitimately relay same-origin in-page widget messages but emit validated=false due to incomplete telemetry instrumentation
  • Developer/test builds of extensions loaded unpacked during QA

Tuning notes: Scope to the two known-vulnerable extension IDs first to cut noise, then expand to any extension forwarding validated=false messages. Allow-list internal/dev extension IDs. Raise severity when the originating tab URL differs from the message origin (cross-origin drive-by).

8 more detection rules (Splunk SPL, Microsoft KQL, Sigma) · unlock with TL Creds. Sign up free to unlock the detection rules or query them with the Threadlinqs MCP server (Purple)
  • Spyder/MaXSS — Over-Broad MV3 Extension Exercising Cross-Origin Privileged Action With No User Gesture (Exploitation) · unlock with TL Creds
  • Spyder — Synthetic Click/Keystroke Injection Into Embedded Web Sessions (Post-Exploitation) · unlock with TL Creds
  • Spyder/MaXSS — AI Conversation Memory & Session Token Exfil to Freshly-Seen Domain (C2/Credential Access) · unlock with TL Creds
  • Spyder/MaXSS — Unvalidated Page-to-Background Message Bridge (Sigma, Exploitation) · unlock with TL Creds
  • Spyder/MaXSS — Hidden Tab to Authenticated Services + Local Collection (Sigma, Post-Exploitation) · unlock with TL Creds
  • Spyder/MaXSS — Screenshot Capture + Web-Service Exfil / Data Manipulation (Sigma, C2/Impact) · unlock with TL Creds
  • Spyder/MaXSS — Extension Opens Hidden Background Tabs to Authenticated Google Services (Post-Exploitation) · unlock with TL Creds
  • Spyder/MaXSS — Programmatic Tab Screenshot Followed by Web-Service Exfil from Extension Context (C2/Exfiltration) · unlock with TL Creds

Indicators of compromise

As of 2026-06-19, Threadlinqs Intelligence tracks 16 indicators of compromise for TL-2026-0870.

Sample indicators (3 of 16)

  • Domain: gemini[.]google[.]com
  • Domain: mail[.]google[.]com
  • Domain: calendar[.]google[.]com

Defanged, TLP:CLEAR. Full set of 16 indicators: unlock with TL Creds.

16 indicators of compromise · unlock with TL Creds. Compare plans

Technical analysis

On 19 June 2026 Rebora Security publicly disclosed two critical vulnerabilities — codenamed 'Spyder' (SiderAI) and 'MaXSS' (MaxAI) — in two AI assistant browser extensions distributed on both Chrome and Edge. The extensions are jointly installed on more than 11 million devices (SiderAI ~10,000,000, extension ID difoiogjjojoaoomphldepapgpbgkhkb; MaxAI ~1,000,000, extension ID mhnlakgilnojmhinhkckjpncpbhabphi). SiderAI is listed among the Chrome Web Store's Top 25 Popular Extensions.

The shared root cause is a broken trust boundary at the content-script message-handling layer. In a Manifest V3 extension, content scripts run in the page's DOM and act as a bridge between untrusted web content and the highly privileged background service worker (which holds the extension's host permissions, tabs, scripting, downloads, and cross-origin fetch capabilities). Both extensions failed to validate the origin/sender of inbound messages: their content scripts accepted sensitive messages that originated from the visited webpage (e.g., via window.postMessage or page-injected DOM events) and forwarded them to the background process as if they were trusted internal commands. Per Rebora: 'MaxAI's content-script made the mistake of accepting such sensitive messages even when they were coming from the webpage' and forwarding them to the background, allowing 'arbitrary websites to force the content-script to ask the background to do just about anything the extension can.'

Spyder abuses SiderAI's design feature of embedding arbitrary websites and invoking user gestures inside those embedded sessions. The attacker 'synthesize[s] an artificial event that activated this functionality from the perspective of a benign webpage,' letting the malicious page simulate clicks and keystrokes across embedded web sessions. This lets a page silently open services such as Google Gemini, drive the AI session, and extract private AI conversation data ('dump the AI's memory of the victim') for external exfiltration.

Demonstrated impact across both flaws: opening hidden/background tabs to victim-authenticated services (Gmail, Google Calendar, Gemini), capturing screenshots of those tabs, simulating clicks and keystrokes, reading email, manipulating documents, exfiltrating AI conversation history, and acting on behalf of the user on virtually any website where the victim is authenticated — enabling account takeover. Because the extensions request broad permissions, Rebora notes a potential path to reading files on the underlying operating system. Exploitation requires no interaction beyond visiting the malicious page, making attacks stealthy and highly scalable.

Rebora attempted responsible disclosure to both vendors and received no response; given the severity it disclosed publicly and notified Google's security teams. No CVE IDs or CVSS scores were assigned or published in the source material at disclosure time. There is no evidence of in-the-wild exploitation reported; the issues are demonstrated by the researchers (PoC-level), and the underlying weakness (CWE-20 improper input validation / CWE-346 origin validation error) is a long-recognized browser-extension anti-pattern.

MITRE ATT&CK techniques by tactic:

Collection

T1005 Data from Local System; T1113 Screen Capture; T1114.002 Remote Email Collection

Execution

T1059.007 JavaScript; T1203 Exploitation for Client Execution

Privilege Escalation

T1068 Exploitation for Privilege Escalation

Command and Control

T1071.001 Web Protocols

Persistence

T1176 Software Extensions

collection

T1185 Browser Session Hijacking

Initial Access

T1189 Drive-by Compromise

Discovery

T1217 Browser Information Discovery

Credential Access

T1528 Steal Application Access Token; T1539 Steal Web Session Cookie

Impact

T1565 Data Manipulation

Exfiltration

T1567 Exfiltration Over Web Service

Remediation

Patches

  • No vendor patch confirmed at disclosure; track SiderAI and MaxAI Chrome Web Store / Edge Add-ons store updates for a version that adds sender/origin validation

Immediate actions

  • Inventory and identify users of SiderAI (extension ID difoiogjjojoaoomphldepapgpbgkhkb) and MaxAI (extension ID mhnlakgilnojmhinhkckjpncpbhabphi) via browser/EDR telemetry
  • Uninstall or force-remove both extensions until vendors ship a fix, given vendors were unresponsive and no patch is confirmed
  • Use enterprise browser policy (ExtensionInstallBlocklist / ExtensionInstallForcelist) to block the two extension IDs in Chrome and Edge

Workarounds

  • Disable the extensions on sensitive accounts/sessions
  • Sign out of high-value Google services (Gmail, Calendar, Gemini) when not in active use to limit hidden-tab abuse
  • Restrict extensions to specific sites only via Chrome's per-extension site access setting (On click / On specific sites)

Longer-term hardening

  • Enforce an extension allow-list and runtime permission review for any extension requesting broad host, tabs, scripting, or downloads permissions
  • Adopt a browser-extension risk-management/EDR capability that monitors content-script-to-background messaging and high-privilege extension actions
  • Educate users that AI assistant extensions run with cross-site privileges and should be treated as high-risk endpoints

Affected and fixed versions

  • Sider.AI — Sider: ChatGPT Sidebar + AI Tools (SiderAI Chrome/Edge extension)
    Vulnerable versions: all versions as of 2026-06-19
    Fixed in: none confirmed
  • MaxAI.me — MaxAI.me (MaxAI Chrome/Edge extension)
    Vulnerable versions: all versions as of 2026-06-19
    Fixed in: none confirmed

Timeline

  • Threadlinqs Intelligence ingested the disclosure as TL-2026-0870 and began detection/remediation analysis.
  • No vendor patch confirmed; extensions remained available in the Chrome Web Store and Edge Add-ons with the vulnerable behavior.
  • No CVE IDs or CVSS scores were assigned or published in the source material at disclosure time.
  • Cyber Security News published secondary coverage describing zero-interaction browser compromise via the two extensions.
  • Researchers demonstrated that a single malicious webpage could, with no user interaction beyond the visit, abuse the unvalidated content-script-to-background message bridge to capture screenshots, open and read hidden tabs, simulate clicks and keystrokes, read Gmail and Google Calendar, and exfiltrate stored AI conversation history across 11M+ combined installs.
  • Rebora Security dubbed the SiderAI content-script message-validation flaw 'Spyder' and the MaxAI counterpart 'MaXSS', characterizing both as improper validation of postMessage/runtime messages forwarded from web page context to the privileged extension background service worker.
  • Rebora informed Google's security teams of the issues given the high severity and scale.
  • Rebora reported that attempts to contact both vendors (SiderAI and MaxAI) about the vulnerabilities failed and received no response.
  • Rebora Security publicly disclosed the 'Spyder' (SiderAI) and 'MaXSS' (MaxAI) Chrome/Edge extension message-validation vulnerabilities affecting 11M+ installs.

Sources cited

Previous and next in vulnerability

Drafted by the Threadlinqs research pipeline from 6 sources; published ; last updated —.

Threadlinqs Intelligence — Real-Time Threat Detection Platform

// threat_feed
$ sort --newest
Showing all threats

Live intelligence console

Threat level

Latest Threats