← Back to the live CVE advisory feed

CVE-2026-76841: Product identification pending

Severity

8.7 (HIGH)

Fixed version

2.12.0

Patch status

Patched

Published

2026-08-24T14:17:01.760

Modified

2026-08-24T14:17:01.760

Why it matters

This matters because the issue affects identity or token handling. Successful exploitation could allow unauthorized authentication, client manipulation, or access to protected services.

Recommended admin actions

• Review and patch based on exposure, asset criticality, and business impact.
• Compare installed versions against the affected version range in the advisory.
• Treat internet-facing systems as higher priority.
• Review logs for suspicious activity related to the affected application or component.

Technical summary

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.

View the official NVD record for CVE-2026-76841

CVE data is sourced from NVD/CNA records and optional enrichment. Validate against the vendor advisory before taking production action.