MD5 Hash Collision in run-llama/llama_index

published 2025-07-10 16 mo ago modified 2026-09-10 1 mo ago also GHSA-5hq9-5r78-2gjh , PYSEC-2026-1566 , PYSEC-2026-2606 , EUVD-2025-20993

cvss: 6.5; version 3.0, CNA
cisa kev: Not listed; not in CISA's catalog
epss: 0.4%; EPSS 0.4% chance of exploitation in 30 days, percentile 30%
public exploits: 0; none known

A vulnerability in the DocugamiReader class of the run-llama/llama_index repository, up to version 0.12.28, involves the use of MD5 hashing to generate IDs for document chunks. This approach leads to hash collisions when structurally distinct chunks contain identical text, resulting in one chunk overwriting another. This can cause loss of semantically or legally important document content, breakage of parent-child chunk hierarchies, and inaccurate or hallucinated responses in AI outputs. The issue is resolved in version 0.3.1.

Patch / fix

fixed in 3 links 1

pypi llama-index 0.12.41
pypi llama-index-readers-docugami 0.3.1
run-llama run-llama/llama_index 0.3.1

github.com/run-llama/llama_index@29b2e07 commit

Vendor advisories

What the vendors themselves say about their products: affected, fixed or not affected, and what to do.

red hat product security vex CVE-2025-6211 updated 2026-06-27 cvss 6.5

llama-index: llama_index MD5 Hash Collision

Red Hat Ansible Automation Platform 2 ansible-automation-platform-24/de-minimal-rhel8, ansible-automation-platform-24/de-minimal-rhel9, ansible-automation-platform-25/de-minimal-rhel8, ansible-automation-platform-25/de-minimal-rhel9, ansible-automation-platform-24/ee-supported-rhel8, ansible-automation-platform-24/ee-supported-rhel9, +5 more
affected: workaround, no fix available “Fix deferred”

Red Hat CSAF VEX © Red Hat, Inc., from https://security.access.redhat.com/data/csaf/v2/vex/; changed: reduced to product statuses and remediations. CC BY 4.0 · csaf

CISA SSVC

Exploitation: Proof of concept is public
Automatable: yes
Technical impact: partial
Assessed: 15 mo ago (2025-07-10)

Exploit prediction

0.36% chance of exploitation in the next 30 days, higher than 30% of all scored CVEs (FIRST EPSS, 2026-11-08).

Exploits and checks

huntr.com https://huntr.com/bounties/1a48a011-a3c5-4979-9ffc-9652280bc389 referenced PoC, tagged "exploit" by a source, unverified

CVSS scores

assigned by	version	score	vector
CNA	3.0	6.5	CVSS:3.0/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:L/A:L
OSV	3.0	6.5	CVSS:3.0/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:L/A:L
GitHub	3.0	6.5	CVSS:3.0/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:L/A:L

Affected packages

ecosystem	package	affected	fixed in
pypi	llama-index	<0.12.41	0.12.41
pypi	llama-index-readers-docugami	<0.3.1	0.3.1

Affected products

run-llama run-llama/llama_index

Classification

Weakness

CWE-440 Expected Behavior Violation

References

github.com /run-llama/llama_index/commit/29b2e07e64ed7d302b1cc058185560b28eaa1352
web, patch
huntr.com /bounties/1a48a011-a3c5-4979-9ffc-9652280bc389
exploit, web, third-party-advisory
nvd.nist.gov /vuln/detail/CVE-2025-6211
advisory
github.com /run-llama/llama_index
package
pypi.org /project/llama-index
package
github.com /advisories/GHSA-5hq9-5r78-2gjh
advisory
pypi.org /project/llama-index-readers-docugami
package

Sources

Fencer Vulns keeps one entry per vulnerability: the CVE id wins, then GHSA, and records that alias each other are merged, except that two CVEs always keep separate entries. The same data is available as JSON.

Summary