Shulou(Shulou.com)06/01 Report--

A few days ago, Tencent found that Google's artificial intelligence learning system TensorFlow has serious security risks, which can be used to pose a security threat. It is reported that this risk is TensorFlow's first security risk of its own, and the pre-research team of Tencent Security platform Department has reported this risk to Google and received thanks.

The pre-research team of Tencent Security platform Department found that * people can generate malicious model files of Tensorflow, * AI researchers, and steal, tamper with or destroy the victims' own AI applications. The risk is very dangerous, on the one hand, the cost is low, ordinary people can implement it; on the other hand, it is so confusing that most AI researchers may be defenseless; at the same time, because of the use of TensorFlow's own mechanism, the latest version of TensorFlow on both PC and mobile will be affected. When the AI model is stolen, it is the painstaking efforts of the developers; but once it is tampered with, the AI is out of control, and the consequences are even more unimaginable.

Google official said that in view of the widespread use of TensorFlow in AI, the TensorFlow website will, like most important software, create a security bulletin page on the site to disclose and fix security vulnerabilities in a timely manner.

It is understood that Google's TensorFlow is widely used by the Internet as an open source framework. According to officials, the framework is used by Google, ebay, airbnb, twitter, uber, Xiaomi, ZTE and other companies. In TensorFlow, AI researchers can reuse models built by others for AI training, or provide AI services directly, which greatly improves the efficiency of development. These public models are usually provided in the form of documents and are generally considered harmless. The security incident has also alerted developers in the field to the potential risks of open source frameworks.

The security pre-research team of Tencent Security platform Department not only focuses on security issues in the frontier areas of AI, Mobile and IoT, but also reports multiple security vulnerabilities to google,apple, Microsoft, adobe and other companies and receives official thanks many times. In addition, they have also studied the safety of drones, radios, the Internet of things and intelligent buildings. In the future, the security pre-research team of Tencent Security platform Department will also continue to devote itself to protecting the security of the Internet and science and technology development.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.