Last Tuesday, September 12, the Senate Judiciary Subcommittee on Privacy, Technology, and Law held hearings on artificial intelligence legislation, following which it was recognized that the successes of the United States government in keeping up with an innovative environment are not significant.

Josh Hawley, a Republican, and a senior member of the aforementioned subcommittee, said that the outsourcing of social networks by Congress to the largest corporations in the world was a disastrous decision.

Richard Blumenthal, Doctor of Technical Sciences, Chairman of the subcommittee, speaks about the need to learn from the existing experience of working with virtual platforms of social interaction within the framework of the legal regulation of artificial intelligence, since, according to him, the loss of control tools in the field of AI will entail significant negative consequences. He stated that the interest of legislators is to create legislation on machine intelligence, and hearings are a means to achieve this goal.

Josh Hawley and Richard Blumenthal presented a one-page structure of artificial intelligence regulation. Their proposals stipulate that AI licensing should be managed by an independent structure. The senators are also convinced that companies engaged in developments in the field of machine intelligence should be legally responsible for the harm caused by their technological systems.

The witnesses at the hearing were Nvidia Chief Scientist and Senior Vice President for Research William Dally, Microsoft Vice Chairman and President Brad Smith, and Boston University law professor specializing in privacy law and technology Woodrow Hartzog.

William Dally characterized the idea of uncontrolled general artificial intelligence as a fictional form of reality from the world of science fiction. He noted that machine intelligence is based on those models that were created by humans. Nvidia’s Senior Vice President for Research also expressed confidence in the companies’ ability to ensure the implementation of a responsible approach to creating powerful and innovative AI tools.

At the same time, William Dally said that no nation or company has the ability to control the situation with the development of artificial intelligence. He noted that AI models are portable, it can be connected to a USB drive and they can be trained in a data center anywhere in the world. According to him, it is possible to regulate the deployment and use of artificial intelligence, but it is impossible to establish comprehensive control over its creation. He also stated his desire to ensure that AI models remain in the United States, and do not end up where the regulatory climate will direct them.

Brad Smith says that it is necessary to ensure the presence of artificial intelligence in the zone of human control so that the risks associated with the use and large-scale spread of advanced technology remain in the space of science fiction and do not become an objective reality of human civilization. He stated that a so-called safety brake is needed, which is similar to an automatic switch in every building, which allows stopping the supply of electricity if necessary. The President of Microsoft shared these arguments in the context of thinking about regulatory measures against companies that use artificial intelligence, for example, to control cars with autopilot or water supply.

Woodrow Hartzog stated that there is no such thing as neutral technology in reality. According to him, it is necessary to apply existing laws, including regulatory legal acts on product liability and consumer protection, regarding artificial intelligence. He is convinced that the United States Government should establish a categorical ban on the extremely dangerous or risky use of AI, especially in the case of such scenarios as biometric tracking, forecasting the actions of police officers, and social assessment. Woodrow Hartzog is convinced that facial recognition technologies, emotions, and biometric data should be in the category of illegal practices prosecuted by law. He noted that in this case, clear measures of struggle are needed, and not procedural protection.

Senator Mazie Hirono, D-HI, asked witnesses how the United States could confirm the fact that foreign governments are using artificial intelligence to create disinformation. Brad Smith, in response to this question, said that Microsoft adheres to the tactics of labeling AI-generated content, but expressed concerns that if such information materials are removed, the technology giant will face accusations of using censorship mechanisms.

Senator John Kennedy, R-LA, asked witnesses about the right of consumers to know about the origin of the content they are viewing, including in cases where its author is an AI. The witnesses replied that it depends on the semantic content of the relevant materials. Woodrow Hartzog stated that as part of the regulation of advanced technology, disclosure of information should be in those situations in which it is effective. If the efficiency criterion is not met, according to him, safety should be ensured. Woodrow Hartzog also stated that if it is impossible to achieve the mentioned goal, the content should not exist at all.

Brad Smith has repeatedly expressed the idea of creating an agency for licensing artificial intelligence in high-risk scenarios. He noted that in order to prevent excessive pressure from the government, it is necessary to take as a sample a model of civil aviation based on industry standards, international coordination, and national regulation.

Currently, the United States authorities are following the concept of a sectoral approach to regulating technological innovations, paying attention to specific risks in specific industries.

As we have reported earlier, China Publishes Final Version of Rules for Use of Generative AI.