PT-2024-37546 · Google+1 · Google Gimini 1.0 Pro+1

Published 2024-08-04 · Updated 2024-08-30 · CVE-2024-6331

Positive Technologies (AI Summary)

CVSS v3.1

7.5

High

Base vector

AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N

Name of the Vulnerable Software and Affected Versions stitionai/devika main branch as of commit cdfb782b0e634b773b10963c8034dc9207ba1f9f stitionai/devika up to version 1.0

Description The issue concerns a Local File Read (LFI) vulnerability by Prompt Injection. It is caused by the integration of Google Gimini 1.0 Pro with specific settings in safety settings, particularly HarmBlockThreshold.BLOCK NONE for HarmCategory.HARM CATEGORY HATE SPEECH and HarmCategory.HARM CATEGORY HARASSMENT, which disables content protection. This allows malicious commands to be executed, enabling the reading of sensitive file contents, such as /etc/passwd.

Recommendations For stitionai/devika main branch as of commit cdfb782b0e634b773b10963c8034dc9207ba1f9f, update to a patched version immediately. For stitionai/devika up to version 1.0, update to a patched version immediately and monitor for exploit attempts. As a temporary workaround, consider disabling the integration of Google Gimini 1.0 Pro with HarmBlockThreshold.BLOCK NONE for HarmCategory.HARM CATEGORY HATE SPEECH and HarmCategory.HARM CATEGORY HARASSMENT in safety settings to re-enable content protection.

Exploit

Fix

Special Elements Injection

Weakness Enumeration

CWE-74

Related Identifiers

CVE-2024-6331

Affected Products

Google Gimini 1.0 Pro

Devika

References · 7

• 🔥 https://huntr.com/bounties/d5ac1051-22fa-42f0-8d82-73267482e60f · Exploit
• https://nvd.nist.gov/vuln/detail/CVE-2024-6331 · Security Note
• https://osv.dev/vulnerability/CVE-2024-6331 · Vendor Advisory
• https://t.me/cvedetector/2405 · Telegram Post
• https://t.me/cvenotify/91414 · Telegram Post
• https://t.me/cvenotify/88872 · Telegram Post
• https://twitter.com/oktsec/status/1823604033199739071 · Twitter Post