(Yicai Global) Dec. 19 -- Blade, the security team of China's Internet colossus Tencent Holdings Ltd. [HK:0700], has found coding with security flaws in TensorFlow, an artificial intelligence development platform of Alphabet Inc. [NASDAQ:GOOGL] (Google), that leaves AI applications edited with this system liable to malicious attacks.

TensorFlow is a programming platform Google provides to AI designers free of charge, Blade's head Yang Yong noted, saying when codes with the security flaw are edited into the AI application, hackers may take over the system by exploiting the vulnerability, steal designers' works, or even infringe users' privacy, as media reported Dec. 17.

Generally, if designers use codes with flaws when programming a robot,  malicious hackers will be able to take over control of this automaton via such defect, Yang explained.

TensorFlow, currently the most widely-used AI development platform, has found application in a multitude of AI scenarios such as speech recognition, computer vision and autonomous driving, and thus the consequences may be unimaginable if control of these falls into the clutches of hackers, Blade said. The team has sent details of the operating mechanism of this flaw to Google, with a request that the latter re-edit the codes.

Industry experts have also expressed their concern over this issue. Enterprises engaging in AI development all integrate algorithms and data, and some information may involve core secrets of enterprises and users, so they will be exposed to high risks if security incidents occur, said an expert with the Shanghai Information Security Trade Association.