US Establishes AI Pre-Review Mechanism: Tech Companies Must Share Models with Government 30 Days in Advance
U.S. President Donald Trump signed an executive order establishing a voluntary framework requiring AI developers like Google and OpenAI to provide the government access to advanced AI models up to 30 days before public release. The order, prompted by safety concerns over Anthropic's Mythos model, represents a compromise between tech companies and the government. It also directs the Treasury, NSA, and CISA to create an AI Cybersecurity Clearinghouse.
📋 Article Processing Timeline
- 📰 Published: June 3, 2026 at 09:14
- 🔍 Collected: June 3, 2026 at 09:23 (9 min after Published)
- 🤖 AI Analyzed: June 6, 2026 at 22:17 (84h 54m after Collected)
(Central News Agency, Washington, 2nd, Combined Foreign Reports) U.S. President Donald Trump today signed an executive order establishing a voluntary framework for artificial intelligence (AI) developers to share advanced models with the government before releasing them to the public.
The core provision of the executive order allows companies such as Google, OpenAI, or Anthropic to provide the government access to their most powerful AI models up to 30 days before their anticipated release.
The order was initiated due to concerns over Anthropic's Mythos model. Because Mythos has the ability to discover and expose vulnerabilities in computer systems, including those of banks, government agencies, and hospitals, Anthropic had not made it publicly available.
The 30-day period is seen as a compromise. The initial draft required the government to have access up to 90 days before release, while tech companies wanted the period shortened to just 14 days.
Google's President of Global Affairs, Kent Walker, stated that the order is "an important development" that will ensure "defenders have the AI tools they need to maintain U.S. security."
The signing is particularly noteworthy as the White House had come close to announcing the measure in recent weeks but had suddenly paused it.
According to reports from Politico and other media outlets, David Sacks, a Silicon Valley venture capitalist who served as the head of AI and cryptocurrency under the Trump administration, called Trump to warn that the measure would slow innovation and weaken U.S. competitiveness in the AI race with China.
Sacks posted on social media platform X last week, saying, "Unnecessary regulation is the biggest threat to American innovation," and that to win the AI race, "we must remove the bureaucratic obstacles set up by 'woke' politicians in state legislatures and Washington."
The executive order also directs the U.S. Treasury Department, the National Security Agency (NSA), and the Cybersecurity and Infrastructure Security Agency (CISA) to jointly establish an "AI Cybersecurity Clearinghouse." This body will work on a voluntary basis with industry and critical infrastructure operators to identify software vulnerabilities and find remediation solutions.
On his first day back in the White House, Trump revoked former President Joe Biden's executive order on AI oversight.
In contrast, the European Union's "AI Act" officially came into effect in 2024, establishing legally binding rules for high-risk AI systems, including mandatory transparency requirements, safety testing for the most powerful models, and incident reporting obligations.
Anthony Aguirre, Executive Director of the Future of Life Institute, which advocates for AI safety, said, "This is an important step in the right direction."
However, he emphasized, "Voluntary frameworks alone are not enough. The government must have the power to prevent the public release of systems that pose an unacceptable risk to national security." (Translation: Xu Ruicheng) 1150603
The core provision of the executive order allows companies such as Google, OpenAI, or Anthropic to provide the government access to their most powerful AI models up to 30 days before their anticipated release.
The order was initiated due to concerns over Anthropic's Mythos model. Because Mythos has the ability to discover and expose vulnerabilities in computer systems, including those of banks, government agencies, and hospitals, Anthropic had not made it publicly available.
The 30-day period is seen as a compromise. The initial draft required the government to have access up to 90 days before release, while tech companies wanted the period shortened to just 14 days.
Google's President of Global Affairs, Kent Walker, stated that the order is "an important development" that will ensure "defenders have the AI tools they need to maintain U.S. security."
The signing is particularly noteworthy as the White House had come close to announcing the measure in recent weeks but had suddenly paused it.
According to reports from Politico and other media outlets, David Sacks, a Silicon Valley venture capitalist who served as the head of AI and cryptocurrency under the Trump administration, called Trump to warn that the measure would slow innovation and weaken U.S. competitiveness in the AI race with China.
Sacks posted on social media platform X last week, saying, "Unnecessary regulation is the biggest threat to American innovation," and that to win the AI race, "we must remove the bureaucratic obstacles set up by 'woke' politicians in state legislatures and Washington."
The executive order also directs the U.S. Treasury Department, the National Security Agency (NSA), and the Cybersecurity and Infrastructure Security Agency (CISA) to jointly establish an "AI Cybersecurity Clearinghouse." This body will work on a voluntary basis with industry and critical infrastructure operators to identify software vulnerabilities and find remediation solutions.
On his first day back in the White House, Trump revoked former President Joe Biden's executive order on AI oversight.
In contrast, the European Union's "AI Act" officially came into effect in 2024, establishing legally binding rules for high-risk AI systems, including mandatory transparency requirements, safety testing for the most powerful models, and incident reporting obligations.
Anthony Aguirre, Executive Director of the Future of Life Institute, which advocates for AI safety, said, "This is an important step in the right direction."
However, he emphasized, "Voluntary frameworks alone are not enough. The government must have the power to prevent the public release of systems that pose an unacceptable risk to national security." (Translation: Xu Ruicheng) 1150603
FAQ
What is the purpose of this executive order?
To provide a voluntary framework that ensures AI model safety without hindering innovation.
Which companies are affected?
Companies developing advanced AI models, such as Google, OpenAI, and Anthropic.
Is this order legally binding?
No, it is a voluntary framework and is not legally binding.