CVE-2023-49797 is a high-severity (CVSS 8.8) CWE-379 vulnerability in pyinstaller. A fix is available for pyinstaller — see the affected versions and patch details below.
Local Privilege Escalation in pyinstaller on Windows
EPSS Exploitation Probability
EPSS (Exploit Prediction Scoring System) is a daily probability model maintained by FIRST.org. It estimates the likelihood a CVE will be exploited in production environments within the next 30 days, derived from real-world threat intelligence signals.
How urgent is this, really
CVE-2023-49797 plotted by exploitation likelihood (EPSS) against impact (CVSS). The shaded corner — EPSS 50%+ and CVSS 7.0+ — is where this CVE doesn't sit, though severity or exploitability alone can still warrant action.
Where this sits among everything scored
Of 377,636 CVEs with a current EPSS score, this one falls in the < 10% band (highlighted). Real counts from FIRST.org, not a sample — log-scaled since the landscape is heavily right-skewed.
Real-World Exposure
pyinstallerReal-time download stats are indexed for npm and PyPI packages. This vulnerability affects PyPI packages — download data is not available via public APIs for these ecosystems.
Description
Impact
A PyInstaller built application, elevated as a privileged process, may be tricked by an unprivileged attacker into deleting files the unprivileged user does not otherwise have access to.
A user is affected if all the following are satisfied:
- The user runs an application containing either
matplotliborwin32com. - The application is ran as administrator (or at least a user with higher privileges than the attacker).
- The user's temporary directory is not locked to that specific user (most likely due to
TMP/TEMPenvironment variables pointing to an unprotected, arbitrary, non default location). - Either:
- The attacker is able to very carefully time the replacement of a temporary file with a symlink. This switch must occur exactly between
shutil.rmtree()'s builtin symlink check and the deletion itself - The application was built with Python 3.7.x or earlier which has no protection against Directory Junctions links
- The attacker is able to very carefully time the replacement of a temporary file with a symlink. This switch must occur exactly between
Patches
The vulnerability has been addressed in https://github.com/pyinstaller/pyinstaller/pull/7827 which corresponds to pyinstaller >= 5.13.1
Workarounds
Is there a way for users to fix or remediate the vulnerability without upgrading?
No workaround, although the attack complexity becomes much higher if the application is built with Python >= 3.8.0.
Affected Packages
| Ecosystem | Package | Vulnerable range | Fix |
|---|---|---|---|
| 🐍PyPI | pyinstaller | all versions | 5.13.1pip install --upgrade 'pyinstaller==5.13.1' |
Detection & mitigation playbook
Open-source dependencyDetect
Scan your dependency tree (package-lock.json, pnpm-lock.yaml, requirements.txt, go.sum, etc.) for pyinstaller, including transitive dependencies — a direct dependency you never call can still pull in a vulnerable version.
Fix
Update pyinstaller to 5.13.1 or later, then make sure no transitive (indirect) dependency still pins the vulnerable range — O3 confirms CVE-2023-49797 is resolved across your whole dependency graph.
Workarounds
If you can't upgrade right away: gate or disable the affected feature, validate untrusted input at the boundary, and avoid passing attacker-controlled data into the vulnerable path. O3's runtime protection blocks exploitation in production as an interim safeguard until the upgrade lands.
How O3 protects you
O3 Security's impact-aware SCA analyses which vulnerable code paths your application actually calls, so a match like CVE-2023-49797 can be triaged on real exposure rather than presence alone.
Tailored to CVE-2023-49797. Runtime protection reduces exposure until a permanent patch is applied and verified — it complements patching, it doesn't replace it.
Frequently Asked Questions
Is CVE-2023-49797 in your dependencies?
O3 Security finds CVE-2023-49797 across PyPI dependencies, including transitive ones, and its impact-aware SCA ranks findings by whether your code actually calls the vulnerable path.