AI Sec Watch

TensorFlow, an open source machine learning platform, had a vulnerability in its `tf.raw_ops.GetSessionTensor` function (a command for retrieving tensor data from a session) where it did not properly validate input arguments, allowing attackers to crash the system through a denial of service attack (making software unavailable by overwhelming or breaking it). The vulnerability was fixed in TensorFlow versions 2.9.0, 2.8.1, 2.7.2, and 2.6.4.

A vulnerability in Oracle Java SE and Oracle GraalVM Enterprise Edition (a high-performance Java runtime) in the JAXP component (Java API for XML Processing, which handles XML data) allows an unauthenticated attacker to partially disable these systems over a network. The vulnerability affects specific versions of Java and can be exploited through untrusted code in web applications or through web services that supply data to the vulnerable APIs, with a severity rating of 5.3 out of 10.

GPT-3 (a large language model that generates realistic human-like text) could be misused by attackers to create convincing phishing attacks (fraudulent messages designed to trick people into revealing sensitive information). The post discusses this threat and mentions that organizations can take countermeasures to protect themselves, though specific details are not provided in the excerpt.

Gradio, a framework for building interactive machine learning demos, has a vulnerability in versions before 2.8.11 where its flagging feature (which saves data to CSV files) can be tricked into storing harmful commands in the file. If someone opens this CSV file in Excel or similar programs, those commands run automatically on their computer.

CVE-2022-0845 is a code injection vulnerability (a flaw where an attacker can insert and execute malicious code) in PyTorch Lightning, a machine learning framework, affecting versions before 1.6.0. The vulnerability stems from improper control over code generation, allowing attackers to run arbitrary code through the affected software.

MLflow, a machine learning platform, had an insecure temporary file vulnerability (CWE-377, a weakness where temporary files are created without proper security protections) in versions before 1.23.1. This vulnerability could potentially allow attackers to access or modify sensitive data stored in temporary files.

TensorFlow (an open source machine learning framework) has a vulnerability where building an XLA compilation cache (a storage system that speeds up machine learning model compilation) with default settings causes a null pointer dereference (a crash that happens when code tries to use a memory location that doesn't exist). This occurs because the default configuration allows all devices, leaving a critical variable empty.

TensorFlow (an open-source machine learning framework) has a vulnerability in its TFG dialect, which is part of MLIR (a compiler framework for optimizing code). An attacker can modify the SavedModel format (the way trained models are saved to disk) to break assumptions the system makes, which can crash the Python interpreter or cause heap OOB (out-of-bounds memory access, where code reads or writes memory it shouldn't).

TensorFlow, an open-source machine learning framework, has a vulnerability in its `simplifyBroadcast` function (a part of the MLIR-TFRT infrastructure, which is the compiler and runtime system) that causes a segfault (a crash from accessing invalid memory) when given scalar shapes (data without dimensions), resulting in a denial of service (making the system unavailable). This affects only TensorFlow version 2.7.0.