CISA KEV alertCyberSIXT
CISA Flags Actively Exploited LiteLLM Flaw Allowing Remote Code Execution
Part of the CVE-2026-40217 story · since 10 Sep
- Thu 10 Sep, 09:14 UTC. The Hacker News, first report. LiteLLM Flaw Exposes AI Gateway Keys and Cloud Credentials. Cites a CVE in CISA KEV.
- Mon 2 Nov, 17:00 UTC. Google Cloud, its first report on the story. Attackers Monetise Stolen Cloud Access Through LLMJacking. Cites a CVE in CISA KEV.
- Tue 3 Nov, 16:30 UTC. The Hacker News. Google reports LLMjacking attacks driving cloud bills to $1 million. Cites a CVE in CISA KEV.
- Wed 4 Nov, 10:01 UTC. SecurityOnline, its first report on the story. Mandiant links UNC6993 to costly enterprise LLM hijacking. Cites a CVE in CISA KEV.
- Fri 6 Nov, 15:30 UTC. SecurityWeek, its first report on the story. Attackers weaponise SAP flaws within days as AI bills soar. Cites a CVE in CISA KEV.
- Wed 18 Nov, 08:01 UTC. SecurityOnline. Google reports autonomous AI agents exploiting flaws in attacks. Cites a CVE in CISA KEV.
- Wed 18 Nov, 19:31 UTC. CISA, its first report on the story. CISA Flags Actively Exploited LiteLLM Flaw Allowing Remote Code Execution. Cites a CVE in CISA KEV. This article, 7th of 8.
- Wed 18 Nov, 22:01 UTC. SecurityOnline. CISA Adds Four Exploited Flaws to Its KEV Catalog. Cites a CVE in CISA KEV.
CISA added CVE-2026-40217 to its Known Exploited Vulnerabilities (KEV) catalogue on 18 November 2026. The vulnerability affects BerriAI LiteLLM and is named the BerriAI LiteLLM Unprotected Alternate Channel Vulnerability; it could let remote attackers execute arbitrary code through bytecode rewriting.
The flaw is an unprotected alternate channel exposed at the `/guardrails/test_custom_code` URI. The available description says attackers can reach it remotely to rewrite bytecode and execute code. CVE-2026-40217 has a CVSS score of 8.8, rated HIGH. A patch is available. Organisations should consult BerriAI’s security advisory and the patch information before applying remediation.
KEV inclusion indicates that CISA has confirmed active exploitation. The supplied data does not identify any known ransomware use. CISA’s remediation deadline is 2 December 2026. The entry does not specify affected LiteLLM versions or provide further details about exploitation, so teams should check the vendor advisory when assessing their deployments.
CISA requires organisations to apply mitigations in accordance with vendor instructions, comply with its BOD 26-04 risk-based security update guidance and Forensics Triage Requirements, and follow applicable BOD 26-04 guidance for cloud services. If mitigations are unavailable, organisations should discontinue use of the product. Federal Civilian Executive Branch (FCEB) agencies are directly subject to CISA’s requirements. All organisations should review their LiteLLM exposure, including internet-facing assets, and prioritise remediation in line with applicable guidance.
For full details, consult the [NVD entry](https://nvd.nist[.]gov/vuln/detail/CVE-2026-40217) and [CISA KEV catalogue](https://www.cisa[.]gov/known-exploited-vulnerabilities-catalog).
How the story developed
8 articles from 5 outlets since 10 Sep. Picked up again on 18 Nov after 12 days quiet. This was the 7th report, 69 days after the first.
- Thu 10 Sep, 09:14 UTC. The Hacker News, first report. LiteLLM Flaw Exposes AI Gateway Keys and Cloud Credentials. Cites a CVE in CISA KEV.
- Mon 2 Nov, 17:00 UTC. Google Cloud, its first report on the story. Attackers Monetise Stolen Cloud Access Through LLMJacking. Cites a CVE in CISA KEV.
- Tue 3 Nov, 16:30 UTC. The Hacker News. Google reports LLMjacking attacks driving cloud bills to $1 million. Cites a CVE in CISA KEV.
- Wed 4 Nov, 10:01 UTC. SecurityOnline, its first report on the story. Mandiant links UNC6993 to costly enterprise LLM hijacking. Cites a CVE in CISA KEV.
- Fri 6 Nov, 15:30 UTC. SecurityWeek, its first report on the story. Attackers weaponise SAP flaws within days as AI bills soar. Cites a CVE in CISA KEV.
- Wed 18 Nov, 08:01 UTC. SecurityOnline. Google reports autonomous AI agents exploiting flaws in attacks. Cites a CVE in CISA KEV.
- Wed 18 Nov, 19:31 UTC. CISA, its first report on the story. CISA Flags Actively Exploited LiteLLM Flaw Allowing Remote Code Execution. Cites a CVE in CISA KEV. This article, 7th of 8.
- Wed 18 Nov, 22:01 UTC. SecurityOnline. CISA Adds Four Exploited Flaws to Its KEV Catalog. Cites a CVE in CISA KEV.
- Known exploited CVE
- No CVE
- Larger: an outlet’s first report
- Square: official source
- Coverage over time
- Quiet time, folded
Show all 8 articles as a listHide the list
- Thu 10 SepDay 01
- LiteLLM Flaw Exposes AI Gateway Keys and Cloud Credentials09:14 · The Hacker News · First report · Exploited CVE · Known exploited CVE
Quiet for 8 weeks
Mon 2 NovDay 531- Attackers Monetise Stolen Cloud Access Through LLMJacking17:00 · Google Cloud · Picked up again · Known exploited CVE
- Tue 3 NovDay 541
- Google reports LLMjacking attacks driving cloud bills to $1 million16:30 · The Hacker News · Known exploited CVE
- Wed 4 NovDay 551
- Mandiant links UNC6993 to costly enterprise LLM hijacking10:01 · SecurityOnline · Known exploited CVE
Quiet for 2 days
Fri 6 NovDay 571- Attackers weaponise SAP flaws within days as AI bills soar15:30 · SecurityWeek · Picked up again · Known exploited CVE
Quiet for 12 days
Wed 18 NovDay 693- Google reports autonomous AI agents exploiting flaws in attacks08:01 · SecurityOnline · Picked up again · Known exploited CVE
- CISA Flags Actively Exploited LiteLLM Flaw Allowing Remote Code Execution19:31 · CISA · official · CISA alert · Known exploited CVE · This article, 7th of 8