The United States is scrambling to strengthen defenses around increasingly powerful artificial intelligence structures before China gains ground.
It may already have time.
New AI models, such as Anthropic’s Claude Mythos and OpenAI’s GPT 5.5-Cyber, have progressed faster than the rules governing what technology can keep up with. Both have demonstrated an uncanny ability to identify software vulnerabilities and launch cyberattacks — skills that hackers and cyber adversaries are eager to exploit.
Recent estimates indicate that the United States has at least six to 12 months before Beijing can acquire a frontier model with capabilities comparable to Mythos or GPT 5.5-Cyber or create an AI competitor that can be used as a cyber weapon.
“It’s a hurricane warning, not a seawall,” Rob T. Lee, chief AI officer at the cybersecurity research firm SANS Institute, said of the time the U.S. should prepare before this new wave of advanced AI designs changes the cybersecurity calculus entirely.
Both Anthropic and OpenAI initially limited testing of their latest AI models to a small group of trusted advocates when the models were first announced in April, as the companies weighed the ramifications of a wider release of the technology. The same month, Chinahe is said to have asked Anthropic for the Mythos casebut he was rejected.
Technology companies,federal agencies,international controlsandallied governmentsthey have been clamoring for the availability of these models in the weeks since, so that they can strengthen their networks.
This race to develop tools to defend against more AI-driven cyberattacks has been accelerated by accusations that China is stealing US technologies to create copycat versions of advanced AI models through distillation attacks, where hackers use the results of a “teacher” model to train their “student” models.
“These models compress what used to take days, weeks or even years of effort into seconds, completely changing the mathematics of cyber security and the multitude of operations defenders must deal with,” said Andrew Rubin, chief executive officer of cyber firm Illumio.
Anthropic has added to the excitement by promising to make “Mythos-level models” available to all of its customers.in the “coming weeks.”An Anthropic spokesperson did not respond to a request for comment as of the time of this publication.
And earlier this week,Anthropic announcedshares Mythos with about 150 new organizations in 15 countries, although it stressed that each group must first meet certain security requirements, without detailing those requirements.
“There’s a certain balance between low availability and wide availability that needs to be struck so that we can collectively find and fix as many critical products as possible before attackers get their hands on these models,” said Lee Klarich, chief product and technology officer at networking company Palo Alto Networks.
An Anthropic spokesperson declined to comment on concerns across China about gaining access to advanced AI prototypes. An OpenAI spokesperson did not respond to a request for comment.
Anthropic last monthpublished a blog postexplaining the company’s views on the competition between the United States and China over AI, where the company said that “it is important that the United States and its allies stay ahead of authoritarian governments such as the Chinese Communist Party.”
“AI will soon be powerful enough to be used to oppress citizens on an unprecedented scale, and even change the balance of power between nations,” Anthropic noted in its blog post.
With this watershed moment for AI fast approaching, the US government is weighing how to support the continued development of American-made technology while balancing the need for greater protection.
The Trump administration has largely taken an approach to restrict cross-border innovation to avoid stifling innovation and remain competitive with China. It was finally motivated to take action after Anthropic warned that the rate of AI developmentthreatening to raise the global economy, public safety and national securityif not secured.
President Donald Trumpsign the executive orderearlier this week that encourages AI companies to submit their powerful new designs to voluntary government review at least 30 days before releasing them to the public.
This call to action has extended to Capitol Hill. Representatives Jay Obernolte (R-Calif.) and Lori Trahan (D-Mass.) unveiled a 269-page draft bill this week that would include regulations on the safety and security of AI, as well as a three-year ban on state laws related to AI. The bill islikely to reach Trump’s deskin its current form because of objections to state legislation, but many still see it as a good first step for Congress.
Whether these measures will be enough to prevent China’s attempts to overtake America’s AI technology remains unclear.
Aaron Rose, a security architect at cybersecurity firm Check Point Software, said that while the United States has advanced in the race for advanced AI, China “is well equipped to develop its own.”
The Chinese government recently made the integration of AI into its economy aa major part of his five-year planpromote its scientific and technological dominance. Trump and Chinese President Xi Jinping toodiscussed AI securityduring their recent meeting in Beijing.
Meanwhile, Chinese AI companies like DeepSeek are thereis reportedly trying to raise billions for “groundbreaking AI research” and compete well with American frontier labsis reportedly seeking a deeper partnershipand the Russian government on AI and cybersecurity, raising concerns among U.S. defense strategists about their efforts toto create a new world orderand powerful technologies at hand.
In a statement, the Chinese Embassy in Washington told POLITICO that “AI is profoundly changing the way people work and live,” and that the Chinese government will host a world AI summit in Shanghai in July.
“It is a new frontier for all mankind,” the embassy adds. “AI should not be owned by a large country, it is still dominated by competition and competition.”
It’s unclear exactly when China will reach America’s frontier AI technology — though it’s certain that it will.
Evan Peña, founder and chief security officer at AI security firm Armadin, suggested that the Chinese government and other groups are likely already using DeepSeek’s latest V4 model to achieve results similar to — and cheaper than — those of Mythos and GPT-5.5-Cyber.
Storiesit has already sniffed thousands of zero-day vulnerabilities– flaws previously unknown to developers – in every major operating system and web browser.
The SANS Institute’s Lee emphasized that defenders should use advanced AI to find and fix vulnerabilities in critical systems before the clock runs out.
“You have time to get up and move what’s important,” Lee said. “But the storm is going to hit, and you need to make sure you’re not just standing outside the field when it happens.”
