# CVE-2026-73847: Emlog CSRF in AI Assistant execute_tool leading to database compromise

Source: https://junglewise.ai/threats/cve-2026-73847-emlog-csrf-in-ai-assistant-execute-tool-leading-to-database  
Publisher: Junglewise Threat Intelligence, https://junglewise.ai/threats  
Published: 2026-08-14. Updated: 2026-09-25  
License: CC BY 4.0 with a link back

CVE-2026-73847 · Severity: medium · CVSS 6.8 · Published 2026-08-14

Emlog is an open-source website building platform with an integrated AI Assistant. The AI Assistant's database query feature lacks CSRF protection and validation, allowing an attacker to trick a logged-in administrator into executing arbitrary SQL queries that can read and modify any database table except the blog content table. An attacker can exploit this to steal administrator password hashes, change user credentials, and gain complete administrative control of the website.

Technologies: [Emlog](https://junglewise.ai/threats/technologies/emlog). Vendors: [Emlog](https://junglewise.ai/threats/vendors/emlog).

## Quick answers

**Is CVE-2026-73847 exploited in the wild?** No. Junglewise Threat Intelligence has no report of CVE-2026-73847 being exploited in the wild.

**How severe is CVE-2026-73847?** Medium severity, with a CVSS score of 6.8, published 14 August 2026.

**Which products and versions are affected?** Emlog 2.6.23 and earlier.

**Is there a fix for CVE-2026-73847?** No fixed version is recorded yet. Check the vendor's advisory in the references below. If you run it, Junglewise drafts a mitigation that fits your setup and applies it once you approve, with senior security engineers behind it.

**How is CVE-2026-73847 attacked?** It is attacked over the network.

**Am I affected by CVE-2026-73847?** Check whether you run Emlog at an affected version. Junglewise does that check for you around the clock: connect your code, cloud and domains, and it matches new vulnerabilities like this one against what you actually run and tells you when one reaches you.

## Technical details

This is a CSRF vulnerability combined with multiple authentication and authorization flaws in the AI Assistant's execute_tool endpoint (admin/ai.php). The vulnerability exploits the absence of CSRF token validation (no LoginAuth::checkToken() call), a missing SameSite attribute on the authentication cookie (enabling Chrome's Lax+POST grace period), and a hardcoded public confirmation string ("confirm") instead of a secret token. Read operations execute with no confirmation requirement, while write operations are protected only by the public string. The password redaction mechanism can be bypassed by aliasing password columns (e.g., SELECT password AS pwd_hash). An attacker delivers a forged cross-site request to a recently logged-in administrator, gaining full database read access (including password hashes) and write access to all tables except blog, enabling direct user table modification for account takeover. No patched version is available as of the advisory date.

## Affected products

- Emlog Emlog 2.6.23 and earlier

## Timeline

- 2026-08-14: disclosed: CVE-2026-73847 and GHSA-v6wr-4x55-7qp5 published

## References

- [https://github.com/emlog/emlog/security/advisories/GHSA-v6wr-4x55-7qp5](https://github.com/emlog/emlog/security/advisories/GHSA-v6wr-4x55-7qp5)

## Related threats

- [Emlog stored XSS in tag names via article editor](https://junglewise.ai/threats/cve-2026-73848-emlog-stored-xss-in-tag-names-via-article-editor)
- [Emlog stored XSS in Markdown article rendering](https://junglewise.ai/threats/cve-2026-53758-emlog-stored-xss-in-markdown-article-rendering)
- [Emlog path traversal in ZIP extraction enables RCE](https://junglewise.ai/threats/cve-2026-53757-emlog-path-traversal-in-zip-extraction-enables-rce)
- [Emlog SQL injection in queryDatabase function](https://junglewise.ai/threats/cve-2026-73850-emlog-sql-injection-in-querydatabase-function)
- [Emlog unauthenticated installation in install.php](https://junglewise.ai/threats/cve-2026-73849-emlog-unauthenticated-installation-in-install-php)