CVE-2026-41486

Ray Data registers custom Arrow extension types (`ray.data.arrow_tensor`, `ray.data.arrow_tensor_v2`, `ray.data.arrow_variable_shaped_tensor`) globally in PyArrow. When PyArrow reads a Parquet file containing one of these extension types, it calls `__arrow_ext_deserialize__` on the field's metadata bytes. Ray's implementation passes these bytes directly to `cloudpickle.loads()`, achieving arbitrary code execution during schema parsing, before any row data is read. In May 2024, Ray fixed a related vulnerability in `PyExtensionType`-based extension types ([issue #41314](https://github.com/ray-project/ray/issues/41314), [PR #45084](https://github.com/ray-project/ray/pull/45084)). In July 2025, [PR #54831](https://github.com/ray-project/ray/pull/54831) introduced `cloudpickle.loads()` into the replacement extension types' deserialization path, reintroducing the same class of vulnerability. ## Impact - **Affected versions**: Ray 2.49.0 through 2.54.0 (latest release as of March 2026). The vulnerable `_deserialize_with_fallback` function with `cloudpickle.loads()` was introduced in commit `f6d21db1a4` ([PR #54831](https://github.com/ray-project/ray/pull/54831), July 2025), first released in Ray 2.49.0. - **Affected configurations**: Any process that uses Ray Data and reads Parquet files. The extension types are registered globally in PyArrow, so all Parquet reads in the process are affected, including `ray.data.read_parquet()`, `pyarrow.parquet.read_table()`, `pandas.read_parquet()`, etc. - **Attacker prerequisites**: The attacker must place a crafted Parquet file where a Ray Data pipeline reads it. No authentication or cluster access is required. The Parquet file must contain a column with a `ray.data.arrow_tensor` (or v2, or variable-shaped) extension type name, which makes this a targeted attack against Ray Data users.

Ray is an AI compute engine. From version 2.54.0 to before version 2.55.0, Ray Data registers custom Arrow extension types (ray.data.arrow_tensor, ray.data.arrow_tensor_v2, ray.data.arrow_variable_shaped_tensor) globally in PyArrow. When PyArrow reads a Parquet file containing one of these extension types, it calls __arrow_ext_deserialize__ on the field's metadata bytes. Ray's implementation passes these bytes directly to cloudpickle.loads(), achieving arbitrary code execution during schema parsing, before any row data is read. This issue has been patched in version 2.55.0.

Ray is an AI compute engine. From version 2.54.0 to before version 2.55.0, Ray Data registers custom Arrow extension types (ray.data.arrow_tensor, ray.data.arrow_tensor_v2, ray.data.arrow_variable_shaped_tensor) globally in PyArrow. When PyArrow reads a Parquet file containing one of these extension types, it calls __arrow_ext_deserialize__ on the field's metadata bytes. Ray's implementation passes these bytes directly to cloudpickle.loads(), achieving arbitrary code execution during schema parsing, before any row data is read. This issue has been patched in version 2.55.0.