Arbitrary SQL via read-only transaction escape
Published Jun 18, 2026 · Updated Jun 18, 2026
SQL injection in pgAdmin 4 9.13 through 9.15 allows remote authenticated users to execute arbitrary SQL through the AI Assistant. The execute_sql_query tool passes LLM-generated multi-statement text to the database, so a leading COMMIT escapes its read-only transaction and later statements run in autocommit mode. An attacker must plant prompt-injection content in data the assistant reads; writes inherit the connected role, and superuser roles enable host command execution.
Summary
What happened
SQL injection in pgAdmin 4 9.13 through 9.15 allows remote authenticated users to execute arbitrary SQL through the AI Assistant. The execute_sql_query tool passes LLM-generated multi-statement text to the database, so a leading COMMIT escapes its read-only transaction and later statements run in autocommit mode. An attacker must plant prompt-injection content in data the assistant reads; writes inherit the connected role, and superuser roles enable host command execution.
The record
- CVE
- CVE-2026-12045
- Published
- Jun 18, 2026
- Updated
- Jun 18, 2026
- Vendor
- pgAdmin Development Team
- Product
- pgAdmin 4
- Classifications
- CWE-77, CWE-89, T1059
- Attack vector
- network
- Privileges
- authenticated
Timeline
How it unfolded
- Jun 18, 2026CVE publishedPublication date reported by the CVE source.
- Jun 18, 2026Record updatedLatest update available in the CVE record.
Exploitability
Present is not the same as exploitable
Compare your product and version with the public record. A matching version still requires validation against your environment.
Is a vulnerable build present?
Compare these published version ranges with your installed build and any vendor patches.
- Affected versionversion=9.13 <9.16
What conditions does exploitation require?
What is affected?
Published CVSS scores
CVSS describes severity. EPSS estimates exploitation probability.
Attacks
What attackers are doing with it
Daily unique IPs observed by Shadowserver honeypots for known exploited vulnerabilities (KEVs). Missing observations do not establish an absence of attacks.
Weakness, pattern, technique
Public exploit references
- AI Assistant transaction-escape regression payloadsproof of concept · demonstrated
Labels summarize the accepted research assessment. They do not indicate a test against your environment.
Technologies
Your stack
See the directory against your own environment.
Your stack
Check the software in your environment
Book a demo to see how Hinoki identifies affected software and validates exploitability in your environment.
Book a demo