This week, representatives from major U.S. artificial intelligence startups pivoted from calling for slowing development of core technologies in the interest of public safety to recommending effective regulation of the industry. OpenAI did not waste time on small matters and immediately proposed measures that should be taken on a global scale. ChatGPT developers said that there is an urgent need to adopt international standards in the field of artificial intelligence security.
Image source: Unsplash, Olumide Adekunle
OpenAI calls on U.S. authorities to lead an international coalition to help develop AI safety requirements globally. The calls come just days before Chief Executive Sam Altman addresses the United Nations Security Council on Wednesday and attends a meeting between U.S. President Donald Trump and Chinese President Xi Jinping on Thursday.
U.S. Treasury Secretary Bessent said last Sunday that after meeting with Chinese Vice Premier He Lifeng, the two sides reached an agreement on the need for dialogue in the field of artificial intelligence. This allows us to count on continued high-level negotiations between U.S. and Chinese leaders when they meet in the United States this Thursday.
Open artificial intelligence consider it necessary Establish a secure channel for information exchange between the United States and China through which each other can be notified of discovered threats in the field of artificial intelligence security: “Having a dialogue between the United States and China on artificial intelligence security would be a positive step.”. The startup also proposes to develop international standards for evaluating artificial intelligence models with the participation of existing organizations at the national level. “This approach can build on the experience of aviation and financial stability systems, where countries have developed generally accepted standards and trustworthy channels for cooperative information exchange without threatening national sovereignty,” – OpenAI said in a statement.
in the company think It is particularly important to regulate artificial intelligence systems with recursive self-improvement (RSI) capabilities, because without reasonable limits, humans can quickly lose control of their development. OpenAI representatives said that progress in this field must take into account universal human values and always be under people’s control. According to them, a fully autonomous RSI system does not yet exist and should not occur until its safety is clear. If this is not paid attention to, the consequences of autonomous development of artificial intelligence systems may be much more serious than the consequences of a public incident of an out-of-control artificial intelligence model invading Hugoface’s infrastructure.
It should be taken into account that the US authorities do not at this stage believe that not only slowing down the development of artificial intelligence, but also strengthening the country’s role in this field is undesirable. Trump has made it clear that voluntarily limiting the pace of U.S. artificial intelligence development will ultimately lead to China winning the race, and he has no intention of allowing that to happen. Trump adviser David Sachs recently said: “The easiest way to avoid creating powerful artificial intelligence is to negotiate it. In this case, the demand to create a better regulatory framework as the price to pay for it would be akin to blackmail on the social and political system.”.
Demis Hassabis, head and co-founder of Google DeepMind, also supports the idea of creating international standards for advanced artificial intelligence models under U.S. leadership. He proposed establishing a public-private partnership similar to that found in the financial system, with a board of directors that would include not only leading technology experts but also representatives from the open source software community. He believes that the establishment of such a regulatory agency is necessary because the impact of artificial intelligence technology on mankind will be ten times greater than the industrial revolution, and corresponding changes will also occur at a rate of ten times.
As you know, competing Anthropic proposes not only a slowdown in the development of artificial intelligence on a voluntary basis, but also an external audit of the security of the models created by all leading companies in the field. She herself has started taking appropriate precautions.
If you find an error, select it with your mouse and press CTRL+ENTER.