← Back to Feed CACHED · 2026-05-17 09:42:19 · cache_key CVE-2025-29912
CVE-2026-7310 · CWE-122 · Disclosed 2026-05-26

A heap-based buffer overflow vulnerability exists in XML parser functionality in the HiDraw

ASSESSED — NOISGATE V0.5
Vendor
Reassessed
Verdict:
01 · The Real Story

This is a booby-trapped blueprint that only goes off after someone is already inside the control room

CVE-2026-7310 is a heap-based buffer overflow in the XML parsing functionality of Hitachi Energy MACH HiDraw. Publicly indexed enrichment for the CVE shows the affected range as HiDraw 9.0 through versions before 9.22, and the attacker path requires a local authenticated user plus a specially crafted XML file that gets opened or imported by HiDraw. The technical upside for an attacker is classic memory-corruption territory: crash, memory corruption, and possible code execution in the context of the engineering workstation user.

In real enterprise conditions this is not a top-tier emergency. The decisive friction is attacker position: this starts *after* the attacker already has access to a niche Windows engineering workstation or can reliably get an engineer to process a malicious project file. That sharply narrows the exposed population, lines up with the very low EPSS, and keeps the practical risk in MEDIUM even though memory corruption always sounds scarier on paper.

"This is a post-compromise engineering-workstation bug, not an internet edge fire."
02 · The Attack Path

4 steps from start to impact.

STEP 01

Land on a HiDraw engineering workstation

The attacker first needs a foothold on a Windows host where MACH HiDraw is installed, or a delivery channel to a user of that workstation. The public CNA description says exploitation requires an *authenticated malicious user with local access*, which means this is not an unauthenticated network entry point. In practice this is a post-initial-access step, not initial access.
Conditions required:
  • Target organization uses MACH HiDraw
  • Attacker has local authenticated access or can operate through a logged-in user session
  • Target host is an engineering workstation rather than a generic server
Where this breaks in practice:
  • HiDraw is a specialized ICS engineering tool with a much smaller deployment footprint than common enterprise software
  • Many HiDraw systems are on segmented engineering networks rather than directly internet-exposed
  • EDR, application control, and workstation hardening often make it harder to establish or maintain local access
Detection/coverage: Asset inventory and software discovery should identify installed HiDraw, but external scanners will not meaningfully detect reachability because the vulnerable surface is local file parsing.
STEP 02

Deliver a malicious XML project artifact

The attacker needs a specially crafted XML file that exercises the vulnerable parser path in HiDraw. No public PoC was found during this review, so weaponization currently appears bespoke rather than copy-paste. The likely delivery mechanisms are shared project files, removable media, internal file shares, or email into the engineer workflow.
Conditions required:
  • Attacker can create a malformed XML file that targets HiDraw parsing
  • Target workflow accepts external or modified project/configuration files
  • Engineer or attacker session can place the file where HiDraw will process it
Where this breaks in practice:
  • No public exploit repository or commodity kit was found
  • HiDraw file-handling workflows may involve change-control gates and known-good project repositories
  • Attachment filtering and sandboxing can catch some obviously malformed file deliveries before they reach engineers
Detection/coverage: Email gateways and sandbox detonation may catch delivery attempts, but vulnerability scanners generally cannot validate exploitability of a local malformed-file parser flaw.
STEP 03

Trigger the vulnerable parser in HiDraw

The malicious XML has to be opened, imported, or otherwise parsed by HiDraw. The CNA CVSS v4 metrics published by Hitachi Energy include UI:A and PR:L, which fits a workflow where a user action is still needed and where low privileges are sufficient once on the box. This is the big downgrade lever: exploitation depends on specific operator behavior on a niche client application.
Conditions required:
  • HiDraw parses the attacker-supplied XML
  • A user action occurs to open/import/process the file
  • The target version is in the vulnerable range
Where this breaks in practice:
  • User interaction is required
  • File preview or review procedures may stop suspicious project artifacts before import
  • Some shops restrict engineering changes to maintenance windows and approved project bundles
Detection/coverage: Good endpoint telemetry may show HiDraw opening unexpected XML from temp, mail, or share paths; there is little signature-based network detection value here.
STEP 04

Corrupt heap memory and gain code execution in user context

If the parser flaw is successfully exercised, the likely outcomes are application crash or code execution in the context of the current user. Because the affected application is an engineering workstation tool, successful exploitation could tamper with project data or stage follow-on actions from a trusted host. But the blast radius is usually bounded to that workstation and whatever privileges that user already has.
Conditions required:
  • The malformed XML reaches the vulnerable parser path
  • Exploit reliability is good enough for the target build and environment
  • Current user context has access to sensitive project files or downstream systems
Where this breaks in practice:
  • Modern exploit mitigations, EDR memory telemetry, and ASLR/DEP can reduce reliability
  • Local user context may be limited and not automatically equal domain or controller compromise
  • No confirmed in-the-wild exploitation evidence was found
Detection/coverage: EDR crash telemetry, child-process anomalies from HiDraw, memory-corruption alerts, and suspicious file writes from the HiDraw process are the best practical signals.
03 · Intelligence Metadata

The supporting signals.

In-the-wild statusNo confirmed in-the-wild exploitation found during review; not listed in CISA KEV.
Proof-of-concept availabilityNo public PoC located across indexed sources reviewed; likely bespoke-only at this stage.
EPSS0.00017 from the user-provided intel block; that is effectively noise-floor territory. *Percentile was not independently verified from a primary EPSS record during this review.*
KEV statusNot KEV-listed as of the CISA KEV catalog page reviewed.
CVSS vector and interpretationNVD displays a Hitachi Energy CNA CVSS v4 vector of CVSS:4.0/AV:L/AC:H/AT:P/PR:L/UI:A/VC:L/VI:L/VA:H/SC:N/SI:N/SA:N, which reads like a local, high-friction, user-assisted flaw rather than remote edge exploitation.
Affected versionsPublic enrichment indexed from MITRE-adjacent sources shows MACH HiDraw 9.0 through before 9.22 as affected.
Fixed versionThe same public enrichment implies 9.22 as the first fixed release. *No distro/backport data was found; this is Windows application software, not a Linux package case.*
Exposure/scanning realityInternet exposure is likely negligible because the vulnerable surface is a local XML parser in a desktop engineering tool, not a network daemon. Shodan/Censys-style scanning is therefore low-value for direct discovery.
Disclosure date2026-05-26 per NVD change history.
Researcher / reporterNot publicly named in the sources reviewed; the CVE source is Hitachi Energy.
04 · The Call

noisgate verdict.

Final Verdict
= UNCHANGED to MEDIUM (4.6/10)

The single most important reason this stays MEDIUM is that the attack requires authenticated local access plus a user-driven file-processing workflow on a niche engineering workstation. That makes it a post-compromise or targeted social-engineering problem with a sharply limited exposed population, not a broad enterprise-edge exploit wave candidate.

HIGH attacker-position requirement is local/authenticated and materially lowers severity
MEDIUM affected version range inferred as 9.0 to before 9.22 from public enrichment
MEDIUM overall severity assessment

Why this verdict

  • Requires local authenticated access: this is not unauthenticated remote code execution; the attacker is already on the box or piggybacking a logged-in engineer session.
  • Requires user-assisted file parsing: the malicious XML still has to be opened/imported in HiDraw, so there is a workflow gate that modern controls and human process can interrupt.
  • Narrow deployment population: MACH HiDraw is a specialized ICS engineering application, which dramatically reduces the number of exposed endpoints compared with mainstream enterprise software.
  • Low threat evidence: EPSS is extremely low, KEV is negative, and no public PoC or active exploitation evidence was found.

Why not higher?

If this were a remotely reachable service or a broadly deployed server component, the memory-corruption angle would justify a much more aggressive score. But the real path is constrained by local authenticated access, user interaction, and a small install base, so the headline technical impact overstates enterprise-wide urgency.

Why not lower?

It is still a real memory-safety bug in an ICS engineering context, and successful exploitation may yield code execution in a trusted workstation used for control-system engineering. That combination keeps it above backlog-only hygiene: even niche engineering endpoints can have outsized operational value.

05 · Compensating Control

What to do — in priority order.

  1. Block untrusted project-file ingestion — Restrict HiDraw XML/project imports to vetted repositories, signed project bundles, or approved transfer paths. For a MEDIUM verdict there is no mitigation SLA, but this is the most effective immediate reduction while you work toward the patch inside the 365-day remediation window.
  2. Tighten workstation application control — Use AppLocker, WDAC, or equivalent to prevent HiDraw-launched child processes and to restrict execution from user-writeable directories. This helps contain a successful parser exploit on Windows engineering workstations; for MEDIUM, there is no mitigation SLA, so apply as part of normal hardening rather than emergency change.
  3. Harden access to engineering workstations — Enforce least privilege, restrict interactive logon, and require MFA or jump-host access where feasible so attackers cannot easily satisfy the local authenticated prerequisite. This matters more than perimeter scanning because the exploit path begins *after* workstation access is obtained.
  4. Monitor HiDraw opening XML from risky paths — Alert on HiDraw reading XML from email cache, downloads, temp folders, removable media, or ad hoc SMB shares. That gives defenders a practical tripwire for exploit delivery attempts on the exact workflow this CVE depends on.
What doesn't work
  • Perimeter firewall changes do not solve this, because the vulnerable surface is a local file parser rather than an inbound network service.
  • Internet vuln scans and external ASM tools do not meaningfully validate exposure here; they cannot see a desktop application's local XML import path.
  • Generic phishing awareness alone is not enough if attackers already have local authenticated access or can stage files through trusted engineering shares.
06 · Verification

Crowdsourced verification payload.

Run this on the target Windows workstation where MACH HiDraw may be installed. Invoke from an elevated PowerShell prompt with powershell -ExecutionPolicy Bypass -File .\check-hidraw-cve-2026-7310.ps1; local admin helps for full registry and file-system coverage, but standard user often works. The script checks uninstall entries and common file locations, then prints VULNERABLE, PATCHED, or UNKNOWN.

noisgate-verify.ps1
POWERSHELLREAD-ONLYSAFE
# check-hidraw-cve-2026-7310.ps1

# Detect probable exposure to CVE-2026-7310 on Windows hosts running Hitachi Energy MACH HiDraw.

# Logic: versions >= 9.0 and < 9.22 are treated as VULNERABLE based on public enrichment.

# Exit codes: 0=PATCHED, 1=VULNERABLE, 2=UNKNOWN


$ErrorActionPreference = 'SilentlyContinue'

function Parse-Version($s) {
    if (-not $s) { return $null }
    $m = [regex]::Match($s, '(\d+(?:\.\d+){0,3})')
    if (-not $m.Success) { return $null }
    try { return [version]$m.Groups[1].Value } catch { return $null }
}

function Test-VulnRange($ver) {
    if ($null -eq $ver) { return $null }
    $lower = [version]'9.0'
    $upper = [version]'9.22'
    if ($ver -ge $lower -and $ver -lt $upper) { return $true }
    if ($ver -ge $upper) { return $false }
    return $false
}

$findings = @()

# 1) Uninstall registry entries

$uninstallRoots = @(
    'HKLM:\SOFTWARE\Microsoft\Windows\CurrentVersion\Uninstall\*',
    'HKLM:\SOFTWARE\WOW6432Node\Microsoft\Windows\CurrentVersion\Uninstall\*',
    'HKCU:\SOFTWARE\Microsoft\Windows\CurrentVersion\Uninstall\*'
)

foreach ($root in $uninstallRoots) {
    Get-ItemProperty $root | ForEach-Object {
        $name = $_.DisplayName
        if ($name -and ($name -match 'HiDraw' -or $name -match 'MACH')) {
            $ver = Parse-Version $_.DisplayVersion
            $findings += [pscustomobject]@{
                Source = 'Registry'
                Name   = $name
                Path   = $_.PSPath
                RawVer = $_.DisplayVersion
                Ver    = $ver
            }
        }
    }
}

# 2) Common install paths

$paths = @(
    'C:\Program Files\Hitachi Energy',
    'C:\Program Files (x86)\Hitachi Energy',
    'C:\Program Files\ABB',
    'C:\Program Files (x86)\ABB'
)

foreach ($base in $paths) {
    if (Test-Path $base) {
        Get-ChildItem -Path $base -Recurse -Include *.exe -File | ForEach-Object {
            if ($_.Name -match 'HiDraw|MACH') {
                $fv = $_.VersionInfo.ProductVersion
                if (-not $fv) { $fv = $_.VersionInfo.FileVersion }
                $ver = Parse-Version $fv
                $findings += [pscustomobject]@{
                    Source = 'File'
                    Name   = $_.Name
                    Path   = $_.FullName
                    RawVer = $fv
                    Ver    = $ver
                }
            }
        }
    }
}

# Deduplicate by path+version

$findings = $findings | Sort-Object Path, RawVer -Unique

if (-not $findings -or $findings.Count -eq 0) {
    Write-Output 'UNKNOWN - HiDraw/MACH installation not found via registry or common paths.'
    exit 2
}

$vuln = $false
$patched = $false
$unknownVer = $false

foreach ($f in $findings) {
    $state = Test-VulnRange $f.Ver
    if ($null -eq $state) {
        $unknownVer = $true
        Write-Output ("UNKNOWN_CANDIDATE - {0} | {1} | version='{2}'" -f $f.Name, $f.Path, $f.RawVer)
    } elseif ($state) {
        $vuln = $true
        Write-Output ("VULNERABLE_CANDIDATE - {0} | {1} | version='{2}'" -f $f.Name, $f.Path, $f.RawVer)
    } else {
        $patched = $true
        Write-Output ("PATCHED_CANDIDATE - {0} | {1} | version='{2}'" -f $f.Name, $f.Path, $f.RawVer)
    }
}

if ($vuln) {
    Write-Output 'VULNERABLE'
    exit 1
}

if ($patched -and -not $vuln) {
    Write-Output 'PATCHED'
    exit 0
}

Write-Output 'UNKNOWN'
exit 2
07 · Bottom Line

If you remember one thing.

TL;DR
Monday morning: identify every Windows engineering workstation with MACH HiDraw installed, confirm whether any are still in the 9.0 to before 9.22 range, and prioritize the ones that handle project imports from shared or less-controlled sources. For this MEDIUM verdict there is no noisgate mitigation SLA — go straight to the 365-day remediation window unless your environment treats engineering workstations as crown-jewel assets; in that case, restrict untrusted XML/project ingestion now and complete the vendor upgrade within the noisgate remediation SLA of ≤365 days.

Sources

  1. NVD CVE-2026-7310 detail
  2. Hitachi Energy cybersecurity alerts and notifications
  3. Public enrichment for affected MACH HiDraw versions
  4. Hitachi Review article describing HiDraw in MACH systems
  5. CISA Known Exploited Vulnerabilities catalog
  6. FIRST EPSS data and report documentation
  7. Tenable CVE record mirroring description and CNA CVSS display
Peer Review

What defenders are saying.

Submit a review attribution: handle + country only
0 flags selected · stored anonymously
Validation Results

Crowdsourced verification outputs.

Results submitted by users who ran the verification payload against their environment.