Threat intelligence company Defused recorded exploitation attempts against CVE-2026-101767 on October 27, 2026, using honeypots that emulate Oracle E-Business Suite installations. The flaw received a CVSS 3.1 score of 9.8, requires no authentication, and was addressed by Oracle seven days earlier in the quarterly Critical Patch Update.
The vulnerability affects the central Application Object Library component and impacts versions 12.2.3 to 12.2.15 of the E-Business Suite. It stems from SQL injection in the Thin Client Framework servlet, reachable over HTTP or HTTPS without valid credentials or user interaction, according to the National Vulnerability Database entry published on October 20.
Chained exploitation extends beyond the database. The injection executes SQL commands under the APPS user account, which holds broad application data access, and this access can be escalated to operating system commands using the application account itself. In practice, attackers can reach financial records, payroll data, and procurement information.
Horizon3.ai reported the issue to Oracle on July 24, 2026, and it remained under coordinated disclosure until the October patch release. The one-week window between patch publication and the first public exploitation attempts on October 27 was classified by Defused as exploratory traffic, with no confirmed successful breaches of customer environments reported as of October 28.
The product’s history increases the assessed risk. Previous critical flaws CVE-2025-61882 and CVE-2026-46817, both scoring 9.8 in the same product, were exploited in campaigns during the past year, with the former used by the Cl0p extortion group. The exposure pattern remains consistent: internet-facing ERP servers with unauthenticated access to the vulnerable endpoint.
Organizations should immediately apply the October 2026 Critical Patch Update to installations running versions 12.2.3 through 12.2.15. Where patching is not immediately feasible, administrators can block the servlet via the E-Business Suite URL firewall, which permits the path by default, as recommended by Horizon3.ai. Detection should focus on unauthenticated POST requests invoking GanttDataServer followed by database queries.


