Skip to main content
Press slash or control plus K to focus the search. Use the arrow keys to navigate results and press enter to open a threat.

Threats Tagged 'ghsa-9cx2-fwcj-qq73'

View all threats tagged with 'ghsa-9cx2-fwcj-qq73'. Filter and sort to focus on specific types of threats.

Pro Console Lifetime

Stop chasing alerts. Route them.

Start free, then upgrade once to turn Radar into an automated delivery engine for your security stack.

Custom feeds / Automations: email, Slack, webhooks, SIEM/MISP / API access (baseline limits)

View Plans & Pricing

API access activates after upgrading in Console -> Billing.

Breach by OffSeqOFFSEQFRIENDS — 25% OFF

Check if your credentials are on the dark web

Instant breach scanning across billions of leaked records. Free tier available.

Scan now

Filter Threats

Narrow down the results by type, severity, or affected countries

Search threats by title, CVE ID, or description. Maximum 100 characters.
Active filters (1):Tag: ghsa-9cx2-fwcj-qq73

Threats Tagged 'ghsa-9cx2-fwcj-qq73'

Click on any threat for detailed analysis and mitigation recommendations

CVE-2026-76841: Improper Control of Generation of Code ('Code Injection') in xorbitsai inferenceCVE-2026-76841
0

Xinference loads models with Hugging Face remote code execution unconditionally enabled, and before version 2.12.0 exposes no setting to disable it. Six loader call sites pass trust_remote_code=True as a literal or as an unconditional default: RerankModel._get_tokenizer in xinference/model/rerank/core.py, SentenceTransformerRerankModel.load in xinference/model/rerank/sentence_transformers/core.py, SentenceTransformerEmbeddingModel.load in xinference/model/embedding/sentence_transformers/core.py, FlagEmbeddingModel.load in xinference/model/embedding/flag/core.py, and two sites in xinference/model/llm/transformers/core.py where PytorchModel._sanitize_model_config and PytorchModel._get_components default the value to True. Because a caller with model launch access can register a model whose type is unknown and supply an arbitrary model path, the server reaches _auto_detect_type and then AutoTokenizer.from_pretrained, which imports and executes Python declared by the model directory's own tokenizer_config.json auto_map, running attacker-supplied code with the privileges of the worker process. Version 2.12.0 gates every site behind allow_trust_remote_code and the XINFERENCE_TRUST_REMOTE_CODE setting, permitting remote code only for bundled built-in models.

Join the discussion

Showing 1 to 1 of 1 result

Filters:Tag: ghsa-9cx2-fwcj-qq73
Page 1 of 1
OffSeq TrainingCredly Certified

Lead Pen Test Professional

Technical5-day eLearningPECB Accredited
View courses