Elon Musk suggested that major artificial intelligence laboratories allow competing companies to test their models before public release. The goal of this initiative would be to enable companies from the United States and China to evaluate each other's systems, searching for potential security vulnerabilities.
For the entrepreneur, this mutual verification among rivals would reduce each company's dependence on conducting its own model evaluation. This suggestion arises in a context of growing alerts about the inherent risks of the technology and calls for greater control over AI advancement.
According to the plan proposed by Musk, the laboratories should facilitate competitors running a series of tests on the models before they are released to the public, involving both American and Chinese companies. Musk stated: 'Instead of correcting your own proof, you would at least have competitors correcting your proof and sounding the alarm if they saw concerns.'
This proposal was presented after leaders of organizations such as Anthropic and OpenAI advocated for moderation in the pace of model development. Researchers have also issued warnings about the potential risks of this technology. Musk and Sam Altman, CEO of OpenAI, expressed support for the suggestion made by Dario Amodei, CEO of Anthropic, regarding the need to slow down development. Amodei had suggested that AI laboratories implement third-party evaluations, with specialists dedicated to verifying security practices.
However, the acceptance of Musk's idea has not yet reached a general consensus. He admitted that his method might not be ideal but argued that it would considerably increase the probability of detecting problems, stating: 'The chances of you finding problems are dramatically higher.'
On the other hand, the Trump administration has shown resistance to requests for stricter limits on AI development. The president posted messages on Truth Social, labeling concerns about the technology as a 'hoax' and a 'farce.' Kevin Hassett, director of the National Economic Council, opined that the private sector is the appropriate environment to manage AI-related issues, although the government maintains monitoring of the industry and can intervene legally if necessary to ensure corporate accountability.
Trump also argues that slowing down AI development could benefit China. Dario Amodei recognized this issue as one of the most complex dilemmas of the debate. On the Chinese side, a spokesperson for the Ministry of Foreign Affairs classified the calls for deceleration as 'alarmism.' Meanwhile, Musk expressed his belief that China would likely embrace his proposal for testing among competitors.