{"id":"CVE-2026-0596","kind":"vulnerability","aliases":["CVE-2026-0596","GHSA-rvhj-8chj-8v3c"],"related":[],"title":"Mlflow: Command Injection when serving models with enable_mlserver=True","description":"A command injection vulnerability exists in mlflow/mlflow when serving a model with `enable_mlserver=True`. The `model_uri` is embedded directly into a shell command executed via `bash -c` without proper sanitization. If the `model_uri` contains shell metacharacters, such as `$()` or backticks, it allows for command substitution and execution of attacker-controlled commands. This vulnerability affects the latest version of mlflow/mlflow and can lead to privilege escalation if a higher-privileged service serves models from a directory writable by lower-privileged users.","summary":"When MLflow serves a model with enable_mlserver=True, the model_uri is placed into a bash -c shell command without sanitization. A model URI containing shell substitutions such as $() or backticks lets an attacker run commands of their choosing.","severity":"critical","cvss":9.6,"epss":0.01393,"exploited":false,"exploited_at":null,"products":[{"name":"mlflow","fixed":["3.9.0"],"affected":"< 3.9.0","ecosystem":"pypi"},{"name":"lfprojects mlflow","fixed":[],"affected":"","ecosystem":"vendor"},{"via":"cpe","name":"mlflow","fixed":[],"affected":"","ecosystem":"pypi"}],"ecosystems":["pypi","vendor"],"tags":["ai-tooling","cwe-78","ml-platform","mlflow","model-serving","python"],"not_affected_if":["fixed-version","mlserver-disabled","trusted-model-uri"],"how_to_check":"- Run `pip show mlflow` and compare the version against `3.9.0`.\n- Search deployment code and scripts for `enable_mlserver=True` or `enable_mlserver`.\n- Check whether model URIs come from users or untrusted sources.","fix":"- `mlflow` `3.9.0`","status":"open","published_at":"2026-03-31T15:16:10+00:00","first_seen_at":"2026-11-16T15:37:43+00:00","triaged_at":"2026-11-16T15:37:51+00:00","updated_at":"2026-11-16T16:08:50+00:00","category":"ai-tooling","confidence":0.8,"triage":{"at":"2026-11-16T15:37:51Z","model":"claude-haiku-5-5","output":{"fix":"- `mlflow` `3.9.0`","reach":"unknown","title":"MLflow: model serving lets crafted model URI run shell commands","impact":"high","summary":"When MLflow serves a model with enable_mlserver=True, the model_uri is placed into a bash -c shell command without sanitization. A model URI containing shell substitutions such as $() or backticks lets an attacker run commands of their choosing.","audience":"developers","category":"ai-tooling","nicknames":[],"who_cares":["mlflow","python","model-serving","ml-platform"],"confidence":0.8,"fix_status":"available","needs_auth":"unknown","worst_case":"Attacker runs arbitrary commands on the model serving host","analyst_note":"","how_to_check":"- Run `pip show mlflow` and compare the version against `3.9.0`.\n- Search deployment code and scripts for `enable_mlserver=True` or `enable_mlserver`.\n- Check whether model URIs come from users or untrusted sources.","plain_action":"","product_name":"","needs_analyst":false,"not_affected_if":["fixed-version","mlserver-disabled","trusted-model-uri"],"who_is_affected":"Teams that serve MLflow models with enable_mlserver=True, where the model_uri can be influenced by a user or comes from a directory that lower-privileged users can write to. A higher-privileged serving service can be escalated through such a directory. Teams on 3.9.0 or later, or not using enable_mlserver, are not affected.","user_interaction":"no","product_ecosystem":"","needs_special_setup":"yes","severity_if_unknown":"unknown"},"cost_usd":0.000746,"input_tokens":4703,"output_tokens":551,"severity_before":"critical"},"analyst_status":null,"analyst_note":null,"search":"'-0596':13A,21A '-2026':12A,20A '-78':30B '8chj':17A '8v3c':18A 'affect':136C 'ai':27B 'ai-tool':26B 'allow':123C 'attack':73B,131C 'attacker-control':130C 'backtick':70B,121C 'bash':55B,106C 'c':56B,107C 'choos':78B 'command':2A,58B,75B,80C,103C,125C,133C 'contain':64B,115C 'control':132C 'cve':11A,19A 'cwe':29B 'direct':99C 'directori':158C 'embed':98C 'enabl':8A,45B,91C 'escal':147C 'execut':104C,128C 'exist':83C 'ghsa':15A 'ghsa-rvhj-8chj-8v3c':14A 'higher':151C 'higher-privileg':150C 'inject':3A,81C 'latest':138C 'lead':144C 'let':71B 'lfprojects':23A 'lower':162C 'lower-privileg':161C 'metacharact':117C 'ml':32B 'ml-platform':31B 'mlflow':1A,22A,24A,25A,34B,40B 'mlflow/mlflow':85C,141C 'mlserver':9A,46B,92C 'model':6A,36B,43B,49B,62B,89C,95C,113C,155C 'model-serv':35B 'place':52B 'platform':33B 'privileg':146C,152C,163C 'proper':109C 'python':38B 'run':74B 'rvhj':16A 'sanit':60B,110C 'serv':5A,37B,41B,87C,154C 'servic':153C 'shell':57B,65B,102C,116C 'substitut':66B,126C 'tool':28B 'true':10A,47B,93C 'uri':50B,63B,96C,114C 'user':164C 'version':139C 'via':105C 'vulner':82C,135C 'without':59B,108C 'writabl':159C","nicknames":[],"mitigations":[],"who_affected":"Teams that serve MLflow models with enable_mlserver=True, where the model_uri can be influenced by a user or comes from a directory that lower-privileged users can write to. A higher-privileged serving service can be escalated through such a directory. Teams on 3.9.0 or later, or not using enable_mlserver, are not affected.","tier":"B","verified":{},"analyst_revisit_at":null,"fix_status":"available","audience":"developers","plain_action":null,"pkeys":["lfprojectsmlflow","mlflow"],"ptokens":["lfprojects","mlflow"],"labels":["aitooling","critical","cwe78","mlflow","mlplatform","modelserving","pypi","python","vendor"],"campaign_type":null,"search_a":"'-0596':13A,21A '-2026':12A,20A '8chj':17A '8v3c':18A 'command':2A 'cve':11A,19A 'enabl':8A 'ghsa':15A 'ghsa-rvhj-8chj-8v3c':14A 'inject':3A 'lfprojects':23A 'mlflow':1A,22A,24A,25A 'mlserver':9A 'model':6A 'rvhj':16A 'serv':5A 'true':10A","search_b":"'-78':30B 'ai':27B 'ai-tool':26B 'attack':73B 'backtick':70B 'bash':55B 'c':56B 'choos':78B 'command':58B,75B 'contain':64B 'cwe':29B 'enabl':45B 'let':71B 'ml':32B 'ml-platform':31B 'mlflow':34B,40B 'mlserver':46B 'model':36B,43B,49B,62B 'model-serv':35B 'place':52B 'platform':33B 'python':38B 'run':74B 'sanit':60B 'serv':37B,41B 'shell':57B,65B 'substitut':66B 'tool':28B 'true':47B 'uri':50B,63B 'without':59B","search_c":"'affect':136C 'allow':123C 'attack':131C 'attacker-control':130C 'backtick':121C 'bash':106C 'c':107C 'command':80C,103C,125C,133C 'contain':115C 'control':132C 'direct':99C 'directori':158C 'embed':98C 'enabl':91C 'escal':147C 'execut':104C,128C 'exist':83C 'higher':151C 'higher-privileg':150C 'inject':81C 'latest':138C 'lead':144C 'lower':162C 'lower-privileg':161C 'metacharact':117C 'mlflow/mlflow':85C,141C 'mlserver':92C 'model':89C,95C,113C,155C 'privileg':146C,152C,163C 'proper':109C 'sanit':110C 'serv':87C,154C 'servic':153C 'shell':102C,116C 'substitut':126C 'true':93C 'uri':96C,114C 'user':164C 'version':139C 'via':105C 'vulner':82C,135C 'without':108C 'writabl':159C","cause_entry":null,"cause_path":null,"disclosures_checked_at":null,"first_public_at":"2026-03-31T15:16:10+00:00","kev":null,"cvss_vector":"CVSS:3.0/AV:A/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H","cvss_by":"GitHub","poc_count":1,"sources":[{"source":"ghsa","external_id":"GHSA-rvhj-8chj-8v3c","url":"https://github.com/advisories/GHSA-rvhj-8chj-8v3c","severity":"critical","published_at":"2026-03-31T15:31:56+00:00","modified_at":"2026-11-16T15:32:52+00:00","first_seen_at":"2026-11-16T15:37:43+00:00","fetched_at":"2026-11-16T15:46:58+00:00","note":null},{"source":"nvd","external_id":"CVE-2026-0596","url":"https://nvd.nist.gov/vuln/detail/CVE-2026-0596","severity":null,"published_at":"2026-03-31T15:16:10+00:00","modified_at":"2026-07-21T19:10:00+00:00","first_seen_at":"2026-11-16T16:08:50+00:00","fetched_at":"2026-11-16T16:08:50+00:00","note":null}],"documents":[{"id":170508,"url":"https://github.com/mlflow/mlflow/pull/19738","title":"Patch: github.com","kind":"other","source":"ghsa","publisher":"vendor","text":null,"published_at":null,"relation":"fix","confidence":null,"note":"Patch: github.com"},{"id":170509,"url":"https://github.com/mlflow/mlflow/commit/202fac4c83ccc8544c087c142b80196d0e60695c","title":"Patch: github.com","kind":"other","source":"ghsa","publisher":"vendor","text":null,"published_at":null,"relation":"fix","confidence":null,"note":"Patch: github.com"},{"id":171138,"url":"https://huntr.com/bounties/2e905add-f9f5-4309-a3db-b17de5981285","title":"Exploit: huntr.com","kind":"poc","source":"nvd","publisher":"research","text":null,"published_at":null,"relation":"poc","confidence":null,"note":"Exploit: huntr.com"}],"disclosures":[],"exploit_sources":[{"type":"poc","url":"https://huntr.com/bounties/2e905add-f9f5-4309-a3db-b17de5981285"},{"type":"epss","score":0.01393,"url":"https://api.first.org/data/v1/epss?cve=CVE-2026-0596"}],"notice":"Free to read, with fair use of about 60 pages a minute. For higher limits or API access: sales@infrastack.ai"}