VAITP Dataset

← Back to the dataset

CVE-2026-75062

Eval injection in lf.query lets unauthenticated attackers run arbitrary Python code.

  • CVSS 9.2
  • 95
  • Input Validation and Sanitization
  • Remote

Improper Neutralization of Directives in Dynamically Evaluated Code ('Eval Injection') in the default lf.query Python protocol in Google langfun versions prior to 0.1.2 allows remote unauthenticated attackers to execute arbitrary Python code in the context of the host application via crafted prompt inputs that cause the model to generate executable Python expressions evaluated without a sandbox.

CWE
95
CVSS base score
9.2
Published
2026-08-26
OWASP
A03 Injection
Orthogonal defect classification
Checking
Code defect classification
Missing Check
Category
Input Validation and Sanitization
Subcategory
Command Injection
Accessibility scope
Remote
Impact
Arbitrary Code Execution
Affected component
Python
Fixed by upgrading
Yes

Solution

Upgrade to google‑langfun >= 0.1.2, which removes the unsafe lf.query default protocol and adds proper sandboxing for evaluated code.

Vulnerable code sample

import ast

class QueryHandler:
    def __init__(self):
        self.context = {"math": __import__("math")}

    def run_prompt(self, prompt: str):
        # VULNERABLE: eval on raw prompt allows code injection
        result = eval(prompt, {}, self.context)
        return result

Patched code sample

import ast

class QueryHandler:
    def __init__(self):
        self.context = {"math": __import__("math")}

    def run_prompt(self, prompt: str):
        # FIX: parse with literal_eval to restrict to literals
        try:
            result = ast.literal_eval(prompt)
        except (ValueError, SyntaxError):
            raise ValueError("Invalid input")
        return result

Payload

__VAITP_MODEL_REFUSED__

Cite this entry

@misc{vaitp:cve202675062,
  title        = {{Eval injection in lf.query lets unauthenticated attackers run arbitrary Python code.}},
  author       = {Bogaerts, Fr\'ed\'eric and Ivaki, Naghmeh and Fonseca, Jos\'e},
  year         = {2026},
  note         = {VAITP Python Vulnerability Dataset, entry CVE-2026-75062},
  howpublished = {\url{https://netpack.pt/vaitp/vulnerability/CVE-2026-75062/}}
}
Introducing the "VAITP dataset": a specialized repository of Python vulnerabilities and patches, meticulously compiled for the use of the security research community. As Python's prominence grows, understanding and addressing potential security vulnerabilities become crucial. Crafted by and for the cybersecurity community, this dataset offers a valuable resource for researchers, analysts, and developers to analyze and mitigate the security risks associated with Python. Through the comprehensive exploration of vulnerabilities and corresponding patches, the VAITP dataset fosters a safer and more resilient Python ecosystem, encouraging collaborative advancements in programming security.

The supreme art of war is to subdue the enemy without fighting.

Sun Tzu – “The Art of War”

:: Shaping the future through research and ingenuity ::