- Countries should adopt regulation as soon as possible
- And developers should take responsibility for fakes and security issues
Microsoft president Brad Smith urged governments to speed up regulation of artificial intelligence.
He considers fakes the biggest problem with neural networks.
“We have to address the deepfake issues. We will have to point out, in particular, that we are concerned about the growth of foreign cyber-influence special operations.
These activities are increasingly being carried out by the governments of Russia, China, Iran.”
By the way, the dipfests have really become quite plausible lately. One such video was shown yesterday by Binance’s director of security.
The hero of the video, generated by a neural network, was the head of the cryptocurrency exchange Changpeng Zhao.
And it is almost impossible to distinguish it from a “living real” person.
But back to Microsoft’s plan. Brad Smith called for regulation not only of AI, but also of manufacturers of such software.
Development companies must commit to protecting physical security, cybersecurity and national security.
Microsoft yesterday proposed a 5-point plan for AI governance.
There are rough rules for regulating neural networks in the public and private sector, possible problems and solutions.
Note that Microsoft is developing its own AI chip codenamed “Athena”.