SINGAPORE, July 7 (Reuters) - Chinese authorities have held meetings with top tech firms over the past month about potentially restricting overseas access to China's most advanced AI models, including those yet to be released, three people familiar with the discussions said.
The talks follow a number of steps by Beijing to keep homegrown AI within the country and underscore how China, like the U.S., is now treating cutting-edge artificial intelligence as a critical national asset that needs controls.
More from Yahoo Scout
Companies present at the talks included tech giants Alibaba and ByteDance as well as startup Z.ai, said the people, who were not authorised to speak to media and declined to be identified.
Since the emergence of DeepSeek's R1 model last year, Chinese AI models have made big inroads globally thanks to their low costs and increasing capabilities. Any decision by Beijing to limit access to those products could ripple across AI markets as costs for many businesses would likely increase.
TOUGHENING UP PENALTIES FOR AI THEFT DISCUSSED
At the meetings, led by China's Ministry of Commerce, participants discussed putting limits on the most advanced AI models — both closed-source and more open versions, according to two of the sources.
Officials talked about making any leak or theft of proprietary AI technology an offence under China's stringent national security law, one of the sources said.
The officials also raised the possibility of implementing new measures to restrict who can fund domestic AI startups, the source added.
The scope of the potential restrictions is still being discussed, two sources said, adding that they may only apply to future models. It was not immediately clear when or even if they would come into force.
China's commerce ministry, which oversees export regulations, and the National Development and Reform Commission — the country's state planning agency whose officials also attended the meetings — did not respond to Reuters requests for comment.
Alibaba, ByteDance and Z.ai also did not respond to Reuters queries.
All three companies have a range of AI models, some closed-source while others are open-weight, meaning users can download, run and customise the underlying systems.
Alibaba's Qwen and ByteDance's Doubao are two of the most widely used AI models in China. Z.ai has recently set Silicon Valley abuzz as the capabilities of its GLM-5.2 model come close to leading U.S. offerings but at a fraction of the cost.
AI MODELS AND NATIONAL SECURITY CONCERNS
U.S. President Donald Trump's administration has also been deeply concerned about national security implications of AI — in particular the potential for American AI products to be misused by military intelligence in China, Russia and other countries of concern.
In June, it ordered that foreign nationals not have access to Anthropic's most advanced Fable and Mythos models, which prompted the company to disable the models for all users globally as nationality could not be verified in real time.
Export controls for Fable, which is designed for the general public, have since been lifted after new safeguards were put in place. But Mythos, designed for cybersecurity professionals, is still only available to some "trusted" U.S. organisations.
Some U.S. AI experts have also said the U.S. needs to regulate the use of Chinese AI models.
ANGST IN CHINA ABOUT MYTHOS THREAT
According to two of the sources, Chinese authorities are deeply worried about the potential for Mythos to exploit software vulnerabilities and that Washington might deploy the model against Chinese interests.
That echoes concerns publicly voiced by state media and Zhou Hongyi, founder of cybersecurity firm 360, a major vendor to government and enterprise clients, who has said China needs to develop its own Mythos.
China this year has implemented numerous measures to protect homegrown AI.
In April, the country's state planner ordered Meta to unwind its $2 billion acquisition of Chinese-founded AI startup Manus. In early June, authorities issued sweeping new rules, tightening control of overseas deals that involve Chinese investors, technology, data and national security.
China had also launched investigations this year into Manus and other local AI startups that had moved abroad, seeking to establish whether they have broken export control laws, according to two of the sources and a third person.
Manus has not responded to requests for comment.
Reuters was not able to learn how any potential new restrictions on overseas access to Chinese AI models might work.
But some hints might be gleaned from a May roundtable of Chinese legal experts on regulations governing open-source AI.
According to a summary of the discussions published in an official Supreme People's Court journal, participants proposed a tiered system: basic open-source tools subject to a simple filing, more advanced technologies facing security reviews, and the most sensitive frontier models barred from public release or restricted to domestic use.
(Reporting by Fanny Potkin in Singapore; Editing by Anne Marie Roantree and Edwina Gibbs)
China's GLM-5.2 matches leading US AI models at a fraction of the cost, threatening hundreds of billions in American AI infrastructure investments.
Government-backed Chinese AI companies carry no capital burden, giving them a structural edge over US rivals racing toward IPOs.
GLM-5.2's open-source release draws cost-sensitive global companies away from expensive US AI products, eroding American market share.
Act now: the analyst who called NVIDIA in 2010 just named his top 10 AI stocks — and Microsoft didn't make the cut. Grab the names FREE today.
OpenAI and Anthropic may want to get their companies public as soon as possible. Microsoft and Meta may want to close financial deals for data centers sometime this year. Several sources say a new AI model from China has already caught up to US AI models across several features.
The model in question is GLM-5.2 from China's Z.ai. Reuters reports that it can "execute complex tasks with minimal prompting." And, as the Chinese have said before, creating GLM-5.2 costs nowhere close to what OpenAI and Anthropic are paying to advance their products. And the Chinese model is open-source, making it widely available. Companies across the world are worried about how much US-made AI products cost them. Some have pulled back their usage because of that.
deepakiqlect / BY-SA 2.0
Chinese AI companies have been accused of pirating parts of US models, but for the time being, they are not being locked out of the US market. However, there are security concerns about using Chinese AI models. And, if the US government blocks them, the adoption rate will become academic.
The news from China represents a tremendous threat for several reasons. One is that American companies are investing hundreds of billions of dollars in development and data centers. If their features fall behind those of Chinese products, what happens to those investments? Are these investments undermined? If so, the financial risk is colossal.
Act now: the analyst who called NVIDIA in 2010 just named his top 10 AI stocks — and Microsoft didn't make the cut. Grab the names FREE today.
Then there is the issue of investors. First, can Antropic and OpenAI get their IPOs to market if their perceived value falls? Additionally, there are the combined market caps of American megatech companies, such as Microsoft (NASDAQ: MSFT), that have bet the house on the future of AI.
Another challenge is that the Chinese AI does not face the burden of the need for capital, at least according to some accounts. The Chinese government provides these companies with capital, either partially or fully, as it has in many other industries, including EVs.
The warnings to the US AI industry are among the oldest expressions of risk. Baseball player Leroy Satchel Paige said, "Don't look back. Something might be gaining on you." And that something could leave parts of the US AI sector in tatters.
Act now: the analyst who called NVIDIA in 2010 just named his top 10 AI stocks — and Microsoft didn't make the cut. Grab the names FREE today.
How Trump’s Anthropic whiplash has helped China ·FastCompany
Mark Sullivan
5 min read
Welcome toAI Decoded, Fast Company's weekly newsletter that breaks down the most important news in the world of AI. You can sign up to receive this newsletter every week via email here.
Trump administration Okays Anthropic's powerful Mythos-class AI models
After placing export restrictions on Anthropic's Claude Fable 5 and Mythos 5 models on June 12—effectively forcing their removal from the market—the Trump Commerce Department has now reversed course. But in the end, the sudden regulatory pivot may play to China's advantage.
Administration officials removed the controls Tuesday evening, after working with Anthropic for two weeks on a refined set of misuse protections. Anthropic promptly announced that it was turning Fable 5 back on for all customers, and turning on Mythos 5 for a select set of approved enterprise customers.
The concern was that foreign actors might use Mythos or Fable to detect and exploit software vulnerabilities in U.S. government or enterprise systems, tasks for which Anthropic had said the models demonstrated surprising capacity. The Commerce Department also asked OpenAI to "stagger" the release of its latest frontier model, GPT-5.6.
The government's intervention rattled investors and tech industry folks because the Trump administration, after pledging to stay away from AI regulation, effectively granted itself a "kill switch" over newly released frontier models.
The government has a legitimate interest in the national security implications of frontier models, but the administration had no ready framework for evaluating the national security risk of new models, or a minimum set of guardrails that AI providers must build in to prevent misuse of the models. Anthropic says it hopes the work it did with the Commerce Department over the past two weeks will lay the groundwork for a set of standards that could apply to all U.S. frontier model providers.
The Trump administration's sudden swerve into AI safety regulation may end up benefiting Chinese model makers. The administration's export restrictions on the Mythos models imposed a pause on the distribution of the U.S.'s best models at a time when Chinese AI labs are rapidly catching up with their U.S. counterparts—and offering an ever-more compelling alternative to expensive models from Anthropic, OpenAI, and the like.
Many enterprises are already adopting open-weight Chinese models from DeepSeek, Alibaba, and others, which they can download for free, refine, and host in their own private clouds. To justify their high per-token costs, the U.S. frontier models must be demonstrably superior to alternative, open-weight models. And they need to be available.
Some enterprises might hesitate to build their AI stack on top of models that are subject to sudden, unscripted restrictions imposed by a regulator. A real-world example: When the U.S. government restricted access to Anthropic's Mythos-class models, the Chinese company Z.ai's (open-weight) GLM-5.2 model quickly climbed leaderboards and gained Silicon Valley users because it was immediately available and less expensive.
U.S. AI companies and their investors are putting hundreds of billions of dollars behind the development of general-purpose frontier models. They're betting that corporations around the world will use their frontier models as the intelligence engines that power major business functions. With U.S. AI policy in flux, and the threat of Chinese models growing, the business case for monolithic, closed AI models seems less viable.
Central bankers are sounding the AI alarm
The AI boom has been framed as a possible solution to weak productivity. At the European Central Bank's annual forum in Sintra, Portugal, this week, central bankers and economists focused on a different question: How much financial risk is building around the technology.
If AI delivers large productivity gains, companies may need fewer workers, which could raise unemployment and weaken consumer spending. If AI disappoints, the capital now flowing into data centers, chips, cloud infrastructure, and the like may fail to produce expected returns. Either outcome could create stress in financial markets and the broader economy.
"If AI overdelivers, it will impact financial stability. If AI underdelivers, it will impact financial stability," Torsten Slok, chief economist at Apollo Global Management, said at the forum, according to Reuters.
The concern is no longer confined to policymakers or investors. A Pew Research Center survey released last year found that 52% of U.S. workers are worried about the future impact of AI in the workplace, while only 36% are hopeful; nearly a third say workplace AI will mean fewer job opportunities for them over the long run. Those fears are landing in an economy where many households have limited room for disruption: The Federal Reserve's latest household survey found that only 63% of adults could cover a $400 emergency expense with cash or its equivalent.
For investors, the question is whether AI companies can justify the valuations and capital spending already attached to them. The worry for workers, though, is whether a technology that raises productivity for firms will also create a labor-market shock.
AI is a difficult problem for central banks because the risks are spread across several parts of the economy. Market corrections, hiring slowdowns, and infrastructure strains can each create their own unique set of pressures. And traditional tools like interest rates and bank supervision may not be well suited to risks that move through technology systems as well as financial markets.
"The leverage on both sides is very worrisome for financial stability," said the International Monetary Fund's Tobias Adrian, according to The Wall Street Journal.
There is precedent for this kind of anxiety. Bank of Canada Governor Tiff Macklem pointed to the dot-com era, when the internet ultimately exceeded expectations while still producing a market bubble. That analogy matters because the dot-com crash was not a verdict against the internet. It was a verdict against timing, valuation, and the assumption that a transformative technology automatically makes every bet on that technology a good one.
The AI cycle may be especially hard to monitor because of its speed and opacity. In markets, AI systems could crowd into similar trades or accelerate bubbles and crashes. In lending, agentic AI could expand credit access while making decisions harder for supervisors to explain. In cybersecurity, advanced models could help companies find vulnerabilities while also giving attackers new tools.
Anthropic's most powerful cybersecurity-focused AI model remains stuck in a regulatory bottleneck, even after the Trump administration dropped export controls on the company's latest AI models. Global users can access Fable 5 starting Wednesday, but Mythos 5 remains limited to selected US organizations through Project Glasswing, while talks continue over broader domestic and international access. The US government is reportedly directing which institutions regain access, leaving foreign governments, companies and financial institutions waiting for clarity.
The episode could be an important signal for investors watching how Washington may regulate frontier AI systems. Anthropic first previewed Mythos in April, initially giving 50 vetted institutions access before expanding Project Glasswing to 150 organizations across 15 countries in June. That rollout was disrupted when the Commerce Department imposed export controls on June 12, requiring Anthropic to obtain US permission before allowing any foreign national to access Fable or Mythos, regardless of location.
The restrictions were introduced after Amazon (NASDAQ:AMZN) security researchers reportedly found a workaround for Fable 5's safeguards, allowing the model to uncover cybersecurity vulnerabilities. The US later eased some restrictions on Mythos and dropped the export controls on Tuesday after Anthropic added new classifiers, or safety filters, to Fable 5. Anthropic is now restoring access to its original US Glasswing partners, but the company said there is no timeline for when international partners may be included, which could keep global AI users and investors watching the policy risk closely.
Analysis-A new, inexpensive Chinese AI model is catching up with Anthropic, OpenAI on their home turf
FILE PHOTO: Illustration shows OpenAI and Anthropic logos ·Reuters
Laurie Chen and Aditya Soni
5 min read
By Laurie Chen and Aditya Soni
BEIJING/BENGALURU, July 2 (Reuters) - Since DeepSeek shocked markets early last year with its cheap but powerful AI model, global consumers have been faced with a choice: Chinese offerings with lower prices and less capability or OpenAI or Anthropic, which have poured billions into development.
A model called GLM-5.2, launched last month by Beijing-based startup Z.ai, may finally be closing that gap in terms of Western interest.
GLM-5.2 has Silicon Valley buzzing with its coding and agent capabilities, or the ability to execute complex tasks with minimal prompting, that almost rival leading U.S. offerings at a fraction of the cost, in what some experts are calling a "mini DeepSeek moment."
It has quickly climbed the usage charts on third-party AI developer platforms like OpenRouter, where it now ranks above Anthropic's models, while executives from cloud data platform Snowflake's CEO Sridhar Ramaswamy to venture capitalist Marc Andreessen have lauded its abilities.
"We now have a Chinese open-weight model that is as good as the currently available models from OpenAI and Anthropic," said David Sacks, U.S. President Donald Trump's former AI czar, last week before Washington lifted curbs on Anthropic's Fable and Mythos models on Tuesday.
Those capabilities have put Z.ai's GLM-5.2 model at the heart of a growing debate about whether China is finally catching up to the U.S. in the AI race, as technology executives warn that Washington's unpredictable regulation of the industry risks hampering its lead in the frontier technology.
"It is just a tick below Opus 4.8 (from Anthropic) and right up there with GPT 5.5 (from OpenAI)," Sacks said of GLM-5.2 on the All-In podcast, adding that "we cannot afford to do things that slow our companies down."
The Anthropic curbs and the delayed public rollout of OpenAI's latest GPT-5.6 model have fueled global demand for the Chinese model, some experts said.
"The international developer community is increasingly aware that relying solely on proprietary, U.S.-based API models carries significant risk," said Brian Tse, founder and CEO of Concordia AI, a Beijing-based consultancy focused on AI safety.
GLM-5.2's positive global reception also suggests increased interest in cheaper open-source development because businesses are getting stung by the rising and often unpredictable costs of using AI to complete tasks, as closed-source agentic AI tools consume more tokens, the units used to measure AI usage.
Z.ai, also known as Zhipu AI, declined to comment. Anthropic and OpenAI did not immediately respond to requests for comment.
GLM-5.2 currently holds fifth place on Artificial Analysis' large language model (LLM) intelligence leaderboard, which ranks performance across a range of benchmarks designed to measure overall capability, including reasoning and coding skills. And it is in the second spot on Code Arena's front-end coding rankings, measuring how well models generate websites and front-end applications, while operating at roughly a sixth of the cost of closed U.S. frontier models like Claude and the GPT series.
Z.ai has not disclosed how much it spent to develop GLM-5.2.
In a reply to Elon Musk on X last month, Z.ai founder Tang Jie said that the Chinese startup could produce a model on par with Anthropic's Fable before the first quarter of next year.
"The shift GLM-5.2 brings is that the open-source model has become a plug-and-play, out-of-the-box product," said Tiezhen Wang, former APAC lead at Hugging Face, a startup that serves as a hub for developers tinkering with open-source models.
"You just deploy the model and without doing any complex fine-tuning systems, it is in a highly usable, ready-to-use state. This drastically lowers the barrier to entry for open-source adoption."
CONVINCING AMERICAN BUSINESSES
One major hurdle to GLM-5.2's large-scale adoption remains data security concerns that have limited use of Chinese models by U.S. enterprises, particularly in regulated industries like banking and cybersecurity. The migration and upgrading of enterprise AI systems typically takes several months, Wang said.
"I have seen some discussion among European companies about whether it could be used in enterprise settings," said Wei Sun, principal AI analyst at Counterpoint Research.
"In the EU and U.S., some clients, partners and regulated industries may simply be unwilling to accept Chinese models in their AI stack, regardless of technical performance or price."
A report earlier this year by non-profit RAND, based on website traffic data across 135 countries, found that Chinese LLMs' global market share jumped to 13% from 3% in the two months after DeepSeek launched its R1 model in January last year. The release sparked a global tech selloff because it contrasted DeepSeek's low cost with massive AI infrastructure spending elsewhere.
China's LLM usage gains were most pronounced in developing countries and those with close political and economic ties to Beijing.
Some experts said concerns about the safety of Chinese AI models were overblown, arguing that running them on U.S. cloud providers or on a company's own servers ensured data security. While major corporations are slow to migrate, tech startups and small- and medium-sized enterprises are moving much faster.
"Developers tend to care less about where a model comes from than whether it works, how much it costs and whether they can deploy or access it reliably," said Poe Zhao, China tech analyst and founder of the Hello China Tech newsletter.
"The likely pattern is partial routing, not overnight replacement of OpenAI or Anthropic. So yes, it is a mini DeepSeek moment but in a narrower, developer-centric sense."
(Reporting by Laurie Chen in Beijing and Aditya Soni in Bengaluru; Editing by Eduardo Baptista and Thomas Derpinghaus)
US in talks with AI companies for voluntary model standards, FT reports
FILE PHOTO: FILE PHOTO: The "AI" acronym is pictured at a tech fair in Paris ·Reuters
Reuters
2 min read
July 1 (Reuters) - The U.S. government is in advanced talks with AI companies to create voluntary standards for the release of new models, with an announcement possible as soon as next week, the Financial Times reported on Wednesday, citing sources.
• Washington has tightened oversight of new model releases to flag risks amid concerns advanced AI could be misused by military intelligence in China, Russia or other countries of concern.
• The standards would set benchmarks for advanced models and timelines, while clarifying who can access them in the United States and abroad, according to the FT report.
• Reuters could not immediately verify the report. The White House, Anthropic, and OpenAI did not immediately respond to Reuters' requests for comment outside regular business hours.
• In June, U.S. President Donald Trump issued an executive order directing agencies to work with leading AI developers to test advanced models before release, and to draft standards for them.
• Google has been in talks with the government ahead of the release of advanced coding models with more sophisticated capabilities, a source told Reuters on Wednesday, adding that the company was also involved in broader discussions on industry standards. FT first reported the details.
• The U.S. Commerce Department on Tuesday lifted export controls on Anthropic's most advanced Fable and Mythos models, less than three weeks after ordering their suspension over national security concerns.
• OpenAI has also faced constraints. Last week, it delayed a full public launch of GPT-5.6 at the U.S. government's request, limiting access to a small group of vetted partners. Both OpenAI and Anthropic are preparing for IPOs.
(Reporting by Carlos Méndez in Mexico City and Preetika Parashuraman in Bengaluru, Additional reporting by Courtney Rozen and Mrinmay Dey; Editing by Maju Samuel and Rashmi Aich)
Amazon's AWS committed $1 billion to AI deployment and ICE launched GPU compute futures, both requiring U.S. frontier models to access global markets.
Minmin Low flagged it remains unclear whether Anthropic's security fixes prevent Chinese rivals from distilling Fable 5 into cheaper competing models.
Act now: the analyst who called NVIDIA in 2010 just named his top 10 AI stocks — and Amazon didn't make the cut. Grab the names FREE today.
Bloomberg reporter Minmin Low told viewers in a recent segment that the U.S. government has removed all foreign-access restrictions on Anthropic's "Fable 5" model, a step she framed as a meaningful competitive reset for American frontier AI labs. In her words, "the race is back on" with Chinese rivals.
Khanthachai C / Shutterstock.com
The U.S. Just Reopened The Global AI Race
According to the segment, the Trump administration had, weeks earlier, required Anthropic to seek permission before allowing foreign access to its most powerful models, Mythos 5 and Fable 5. Low reported that restrictions on the more powerful Mythos 5 were lifted only for vetted companies, such as critical infrastructure providers and cyber defenders. For Fable 5, a less powerful, public-facing model intended for widespread use, all foreign access restrictions are being removed.
Low framed the outcome as a major near-term win for Anthropic, which could lift the company's valuation ahead of its planned IPO.
Why This Matters Beyond Anthropic
Low's argument, as presented in the segment, is that keeping the restrictions in place would sharply limit international deployment of U.S. frontier models from Anthropic, OpenAI, and other labs. These restrictions would make it more difficult for the US to deliver on its goal of remaining the international leader in AI.
Act now: the analyst who called NVIDIA in 2010 just named his top 10 AI stocks — and Amazon didn't make the cut. Grab the names FREE today.
The easing of these restrictions lands as U.S. hyperscalers are pouring capital into AI deployment infrastructure. Amazon (NASDAQ:AMZN)'s AWS is investing $1 billion in a new Forward Deployed Engineering unit designed to help customers operationalize AI systems, including a managed service for classified defense workloads. Separately, Intercontinental Exchange (NYSE:ICE) and NATIVX are launching energy-normalized GPU compute futures based on the COIL Index, a sign that compute access is being treated as a tradable strategic asset. Both moves assume U.S. models can actually reach global customers.
Anthropic itself is showing up in enterprise security workflows. Cybersecurity firm Tenable (NASDAQ:TENB) disclosed partnerships with Anthropic's Project Glasswing and OpenAI's Daybreak Cyber Partner Program, and was named the current company to beat for AI-powered exposure assessment in a June 2026 Gartner report. Wider availability of Fable 5 would increase the surface area for those kinds of integrations.
The Security Risk Hasn't Gone Away
Low flagged the national security tension directly in the segment, noting that U.S. hyperscalers have accused Chinese competitors of "distillation," generating large amounts of data from frontier models to train their own cheaper systems. She said it remains unclear whether Anthropic's security fixes truly prevent Fable 5 from being distilled by Chinese players.
Low cited Chinese Premier Li Keqiang as saying you cannot "copy your way to an innovation edge," arguing that long-term AI leadership will ultimately depend on original innovation rather than imitation.
The Bottom Line
The Trump Administration's decision removes a major hurdle for Anthropic's global expansion and gives the company greater freedom to deploy its public-facing AI model overseas. If that translates into broader enterprise adoption, it could strengthen Anthropic's business ahead of a potential IPO while also benefiting partners across the U.S. AI ecosystem.
The bigger question is whether the security measures Anthropic put in place are enough to prevent Chinese competitors from using distillation techniques to build rival models more cheaply. That balance between expanding U.S. AI exports and protecting America's technological edge will likely remain one of the defining policy debates as the AI race continues to accelerate.
Act now: the analyst who called NVIDIA in 2010 just named his top 10 AI stocks — and Amazon didn't make the cut. Grab the names FREE today.
Anthropic’s AI models are back online after a two-week government standoff—settling the company and administration into a fragile truce
Anthropic CEO Dario Amodei ·Fortune·Ruhani Kaur—Bloomberg/Getty Images
Tristan Bove
4 min read
Anthropic has restored global access to Fable 5, two weeks after the Trump administration slapped export controls on the company's most powerful AI model, citing security concerns.
The AI company announced Tuesday the controls had been relaxed—a shift also confirmed by U.S. Commerce Secretary Howard Lutnick, who wrote on X the government had spent the past two weeks working with Anthropic to ensure Fable had "alignment" with U.S. interests.
Access to Fable, as well as to a related model produced by Anthropic called Mythos 5, has been restricted since mid-June. The U.S. government first banned the models' use by foreign nationals, but because the ban applied to users regardless of whether they were inside or outside the U.S., and would have also applied to foreign Anthropic employees, the company quickly announced it would be suspending access for all users.
The Trump administration said the decision had been taken on national security grounds, fearing the possibility hostile actors could "jailbreak" the models and bypass the built-in guardrails designed to prevent malicious use.
But those restrictions turned out to be short-lived. Last week, the government allowed Anthropic to release its Mythos model to a select group of over 100 U.S.-based companies and federal agencies, partially undoing the restrictions. The move mirrored Anthropic's own strategy when the powerful models were first released. The company first unveiled Mythos to an exclusive group of users earlier this year, before rolling out Fable, a version with the same underlying capabilities but with more guardrails and safety filters, designed for general use.
Now that both Mythos and Fable are available again to their respective users, Anthropic has effectively reset the frontier AI model landscape to the same status quo it left behind two weeks ago before the government intervened, although the company's announcement on Tuesday suggested it would work more closely with the White House moving forward.
Anthropic, one of the world's leading AI startups, with a private-market valuation of roughly $1 trillion, said it would "scale up our government collaboration" and work to develop a "shared industry framework" with the administration and the private sector. The company specifically said it would be working with the likes of Amazon, Microsoft, and Google to set industry standards of how to assess potential vulnerabilities in frontier models. This system would see security researchers and industry partners using a common scoring system of jailbreak severity, and a shared playbook for how to respond.
As for the government, Anthropic's announcement suggested it will seek to preemptively work through national security concerns and avoid future release delays as happened with Mythos and Fable. The company said it would prerelease frontier models for federal authorities to review and test for weaknesses, and implement dedicated research teams of Anthropic employees to ensure its models would not run afoul of government priorities.
"Our hope is that this collaboration, along with our proposed consensus industry framework, will serve as the basis for systematic rules for the whole industry—and even offer the beginning of a template for effective global coordination on the risks and benefits of AI," the company wrote.
The announcement represents a reset of sorts for Anthropic's turbulent relationship with the federal government. The Department of Defense designated the company a "supply-chain risk" earlier this year, also on national security grounds, after Anthropic refused to allow the Pentagon use of its models for surveillance or autonomous warfare operations.
The company and the government have since seemed to settle into an uneasy truce. Unlike other prominent tech CEOs, Anthropic's boss, Dario Amodei, has been resistant to several of the Trump administration's demands, including, according to Politico, requests that the company voluntarily pull Fable off the market.
Anthropic's willingness to work more closely with the government may signal a thawing. The company still relies heavily on federal contracts, but Anthropic's incentives to appease the Trump administration likely go beyond its direct business dealings with the government. The San Francisco company, which was valued at $965 billion in its most recent funding round in May, recently submitted SEC paperwork for a highly anticipated initial public offering. Anthropic is in a tight race with rival OpenAI to list shares and tap public market investors, following the blockbuster SpaceX IPO. Open hostility with the Trump administration could stand in the way of Anthropic's IPO plans, as would the overhanging risk that the government could step in to turn Anthropic's models off whenever it sees fit.
