VulnerabilityPublished 20 Sep 2026

CVE-2026-94093: DLR-RM stable-baselines3 · Deserialization

Affects DLR-RM stable-baselines3Source CVE Database V5

Compare

Public exploit available · CVSS 5.3 Medium · patch available : patch DLR-RM stable-baselines3 as a priority.

Data as of 5 Nov 2026, 00:01 UTC

CVSS 4.0 base score

5.3Medium

How severe the flaw is on the 0–10 CVSS scale.

EPSS: 0.4% chance of exploitation in 30 days – higher than 39% of CVEsExploit Prediction Scoring System: the likelihood attackers will use this flaw soon.
Patch: Available

CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:P/VC:L/VI:L/VA:L/SC:N/SI:N/SA:N/E:P

Base metrics

Attack Vector: Network
Attack Complexity: Low
Attack Requirements: None
Privileges Required: None
User Interaction: Passive
Vulnerable system confidentiality: Low
Vulnerable system integrity: Low
Vulnerable system availability: Low
Subsequent system confidentiality: None
Subsequent system integrity: None
Subsequent system availability: None

Threat metrics

Exploit Maturity: POC

Bold values are the worst case: easiest to exploit or largest impact.

Attack Vector: Network: The vulnerable system is bound to the network stack and the set of possible attackers extends beyond the other options listed below, up to and including the entire Internet. Such a vulnerability is often termed “remotely exploitable” and can be thought of as an attack being exploitable at the protocol level one or more network hops away (e.g., across one or more routers).
Attack Complexity: Low: The attacker must take no measurable action to exploit the vulnerability. The attack requires no target-specific circumvention to exploit the vulnerability. An attacker can expect repeatable success against the vulnerable system.
Attack Requirements: None: The successful attack does not depend on the deployment and execution conditions of the vulnerable system. The attacker can expect to be able to reach the vulnerability and execute the exploit under all or most instances of the vulnerability.
Privileges Required: None: The attacker is unauthorized prior to attack, and therefore does not require any access to settings or files of the vulnerable system to carry out an attack.
User Interaction: Passive: Successful exploitation of this vulnerability requires limited interaction by the targeted user with the vulnerable system and the attacker’s payload. These interactions would be considered involuntary and do not require that the user actively subvert protections built into the vulnerable system.
Vulnerable system confidentiality: Low: There is some loss of confidentiality. Access to some restricted information is obtained, but the attacker does not have control over what information is obtained, or the amount or kind of loss is limited. The information disclosure does not cause a direct, serious loss to the Vulnerable System.
Vulnerable system integrity: Low: Modification of data is possible, but the attacker does not have control over the consequence of a modification, or the amount of modification is limited. The data modification does not have a direct, serious impact to the Vulnerable System.
Vulnerable system availability: Low: Performance is reduced or there are interruptions in resource availability. Even if repeated exploitation of the vulnerability is possible, the attacker does not have the ability to completely deny service to legitimate users. The resources in the Vulnerable System are either partially available all of the time, or fully available only some of the time, but overall there is no direct, serious consequence to the Vulnerable System.
Subsequent system confidentiality: None: There is no loss of confidentiality within the Subsequent System or all confidentiality impact is constrained to the Vulnerable System.
Subsequent system integrity: None: There is no loss of integrity within the Subsequent System or all integrity impact is constrained to the Vulnerable System.
Subsequent system availability: None: There is no impact to availability within the Subsequent System or all availability impact is constrained to the Vulnerable System.
Exploit Maturity: POC: Based on threat intelligence sources each of the following must apply: · Proof-of-concept is publicly available · No knowledge of reported attempts to exploit this vulnerability · No knowledge of publicly available solutions used to simplify attempts to exploit the vulnerability

Definitions: FIRST CVSS v4.0 specification.

Exploitation evidence

Disclosed: evidence found.
CVE published 20 Sep 2026
Detectable: no evidence on record.
No public nuclei template indexed
Weaponized: evidence found.
CVSS exploit maturity: proof of concept public
Exploited in the wildExploited: no evidence on record.
Not in CISA KEV and no source reports exploitation
Observed by OffSeq MirageSeen by Mirage: no evidence on record.
No attack traffic recorded by OffSeq Mirage sensors

A missing step is not evidence of safety. Sources: cve.org (opens in a new tab), CVSS vector, the Metasploit, Exploit-DB and nuclei-templates index as of 5 Nov 2026.

Disclosed: The CVE record is public on cve.org. Its publication is t = 0.
Detectable: A public nuclei template can find exposed instances, for defenders and attackers alike.
Weaponized: A Metasploit exploit module or Exploit-DB entry exists (Metasploit scanner modules do not count), or CISA SSVC or the CVSS exploit maturity reports a public proof of concept.
Exploited in the wild: CISA KEV lists it, CISA SSVC rates exploitation active, or a source reports attacks.
Observed by OffSeq Mirage: OffSeq Mirage honeypot sensors recorded attack traffic for this CVE, or found exposed hosts without attack traffic.

Moments are spaced evenly, not to scale; a dashed line marks a gap of six months or more. Deltas count from t = 0. Sources that give only a date (CISA KEV, Exploit-DB, or a time of exactly 00:00Z) read “same day” rather than hours. Metasploit modules carry the flaw’s disclosure date, not their release date, and nuclei templates carry no date, so both appear above but not on the timeline.

What to do

Affected software

Product	Affected	Fixed in
DLR-RMstable-baselines3	Affected 2.0 – 2.9.0 10 affected builds	Fixed in–

CPE configurations (1)

cpe:2.3:a:dlr-rm:stable-baselines3:*:*:*:*:*:*:*:*

AI-suggested

A patch is available for this vulnerability. Users should update stable-baselines3 to the fixed version once the outstanding hardening (PR #2264) is released and integrated. Until then, avoid loading untrusted data with the affected load APIs (PPO.load, load_replay_buffer, VecNormalize.load). Monitor the vendor advisory for the official fix and apply it promptly.

Description

CVE-2026-94093 is a medium severity deserialization vulnerability in DLR-RM stable-baselines3 affecting versions 2.0 through 2.9.0. It involves unsafe pickle deserialization in the core load APIs (PPO.load, load_replay_buffer, VecNormalize.load) within save_util.py. Although a partial hardening was introduced in version 2.9.0, it was reverted in later development branches, leaving the vulnerability exploitable. The flaw allows remote attackers to manipulate deserialization, potentially leading to code execution or other impacts. A patch is available but not yet fully integrated into all branches. No known exploits are currently observed in the wild.

Impact

The vulnerability allows remote attackers to exploit unsafe pickle deserialization in stable-baselines3 load functions, potentially leading to arbitrary code execution or other malicious actions during model loading. The impact is rated medium severity with a CVSS 4.0 score of 5.3. There are no confirmed reports of exploitation in the wild at this time.

Summary

The vulnerability CVE-2026-94093 affects stable-baselines3 up to version 2.9.0 in the functions PPO.load, load_replay_buffer, and VecNormalize.load located in save_util.py. These functions perform deserialization using pickle without sufficient safeguards, enabling remote attackers to manipulate the deserialization process. Although version 2.9.0 introduced a hardened PyTorch tensor load path using weights_only=True, this was reverted in the master branch to maintain compatibility with PyTorch 1.13, thus removing the hardening. Previous fixes only gated loading in the huggingface_sb3 package, leaving the core stable_baselines3 load APIs vulnerable. The vulnerability has a CVSS 4.0 base score of 5.3 (medium severity) and has been publicly disclosed. A patch is available but pending full deployment.

References

Community reviews

Loading reviews…