AI industry leaders have expressed concern that antitrust laws could hinder the coordination of AI development and slow down efforts to ensure safety. according to wired OpenAI recently asked U.S. lawmakers to clarify whether organizing an industry-wide AI slowdown is legal, citing sources.
Image source: Growtika/unsplash.com
OpenAI chief scientist Jakub Pachocki recently said there is a need to slow down. He said this would make it possible to build reliable control mechanisms on self-learning artificial intelligence systems. However, experts say the proposed coordinated action to slow the future development of artificial intelligence could violate U.S. law, specifically the Sherman Antitrust Act.
In July, a bipartisan group of lawmakers introduced the Cooperation to Counter Hostile Threats and Security Risks Act, which would protect artificial intelligence companies from coordinating to slow down antitrust challenges. The document is currently awaiting consideration by the Legal Committee.
Experts, meanwhile, believe that mentioning antitrust risks is just a convenient cover to avoid discussing the real reasons why AI developers are avoiding collaboration.
In fact, there is intense competition between companies for some of the emerging markets for advanced models. There are also concerns that China’s progress in artificial intelligence will have a significant impact on national security. And, on top of that, AI developers have wildly different ideas about the best way to develop safe AI, so they may be unwilling to collaborate with each other on the issue, Wired writes.
“The first step is for industry leaders OpenAI and Anthropic to end their feud and work together on a new proposal. – OpenAI co-founder John Schulman said on social network X, – They cited antitrust laws, but that was a lie—antitrust laws prohibit certain agreements but not joint proposals. “
Debate has intensified this summer over the risks posed by continued competition among artificial intelligence companies to create and release more powerful models. Adding fuel to the fire are statements from former Anthropic and OpenAI researcher Jacob Coxon, who said artificial intelligence developers are putting humanity at risk.
In addition, a series of security incidents, including the Hugging Face hack of the OpenAI agent, shows that industry protection measures have not kept pace with the growth of model capabilities.
If you find an error, select it with your mouse and press CTRL+ENTER.