$32B Z.AI Founders’ Project Got Funds From MI5-Flagged Body

Four founders of Z.AI, the Chinese AI company also known as Zhipu, worked on a research project co-funded by an institute that the UK’s MI5 says has “very strong ties” to China’s Ministry of State Security, Bloomberg reported on 19 October 2026. Bloomberg based the report on publicly available Chinese research papers and MI5’s espionage alert.

The institute is the Beijing-based China General Technology Research Institute, or CGTRI. Bloomberg wrote that the MI5 alert didn’t mention Z.AI, and that it’s not clear what, if anything, any of the founders may have known about CGTRI’s alleged relationship with Chinese intelligence. China’s embassy in the UK called MI5’s allegations against CGTRI “baseless.”

Business Pill · SOVEREIGN AI

A short explainer on sovereign AI, the idea of countries building and controlling their own AI. It teaches the general idea only and says nothing about any company or government in this story.

What Bloomberg Found

The project ran from January 2018 through December 2021 and had a total approved budget of CNY2.94 million ($437,000 at the time), according to a 78-page report its leaders compiled, which Bloomberg described as previously unreported. The report was published on a Chinese government website that provides research funding records, Bloomberg said.

It was co-funded by CGTRI and the National Natural Science Foundation of China, and jointly conducted by CGTRI, Tsinghua University and Southeast University, according to the researchers’ report as cited by Bloomberg. Neither university responded to requests for comment, Bloomberg said.

The project sought to develop AI tools to analyze social media posts, spot key events and map peoples’ connections, Bloomberg reported. A key goal was a knowledge base of “China-related key persons,” according to the report, which didn’t elaborate on who such people would be.

What Was Posted on X

Bloomberg reporter Kate O’Keeffe posted the report, written with Maggie Eastland, at 23:10 UTC on 19 October. Her post says four founders of Z.AI worked on research in a joint project with an institute British intelligence accused of having “very strong ties” to China’s main spy agency. As Bloomberg wrote, MI5’s alert does not mention Z.AI, and the allegations about the institute are MI5’s.

Co-author Maggie Eastland quoted the post at 23:41 UTC, describing the report as covering the early days of Z.AI and how four of its founders worked on an overlapping research project funded by CGTRI, a group the UK says has “very strong ties” to China’s top spy agency. Z.AI didn’t respond to Bloomberg’s requests for comment, according to Bloomberg.

The Four Founders Named

Z.AI co-founder Li Juanzi, a professor at Tsinghua University, led the project alongside CGTRI’s Cao Weibing and a professor from Southeast University, the researchers’ report says, according to Bloomberg.

The report lists three other founders, Tang Jie, Xu Bin and Zhang Peng, who is Z.AI’s chief executive officer, as co-authors of papers, co-inventors on patents or holders of software copyrights in its directory of research achievements, Bloomberg reported. The report explicitly states that nonparticipants can’t be included, Bloomberg said.

Z.AI’s interim financial report, filed last month, lists five founders as “key persons,” Bloomberg wrote. The only one not named in the project’s documentation is Chairman Liu Debing.

What MI5 Said About CGTRI

On 30 September 2026, MI5 issued a Security Service Espionage Alert stating that the primary purpose of CGTRI “is to fund research that directly improves Chinese Ministry of State Security (MSS) technical capability for espionage,” according to MI5’s published notice.

MI5 says CGTRI funds academic research in China on topics including AI and cybersecurity, and that “more than 100 UK-linked academics have contributed to research projects funded by MSS via CGTRI.” It adds that “in some cases, academics may not be aware that CGTRI is funding the Chinese research project they are contributing to.”

The notice says the alert “puts the fact that CGTRI has very strong ties to MSS in the public domain” and that “any institution or individual continuing to conduct research ultimately funded by CGTRI should take their own independent legal advice.”

The Responses

Z.AI didn’t respond to emailed requests for comment, nor did the four founders, Bloomberg reported. A CGTRI representative said by phone that the institute doesn’t do any external communication with anyone, according to Bloomberg.

Liu Chang, a spokesperson for the Chinese Embassy in Washington, told Bloomberg: “I’m not familiar with the specifics.” He added: “We firmly oppose attempts to throw mud at China by distorting facts.” A White House spokesperson referred questions to the Commerce Department, which declined to comment, Bloomberg said.

James Mulvenon of Pamir Consulting, who reviewed the report at Bloomberg’s request, said China’s Ministry of State Security “had an obvious interest in funding this research and reaping the targeting benefits.”

Z.AI’s Existing US Restrictions

The US Commerce Department’s Bureau of Industry and Security added Beijing Zhipu Huazhang Technology and related Zhipu entities to the Entity List in a rule published in the Federal Register on 16 January 2025. The rule says “these entities advance the People’s Republic of China’s military modernization through the development and integration of advanced artificial intelligence research.”

Z.AI previously said the US decision to restrict its access to technology “lacked factual basis” and wouldn’t have a substantial impact on its business, Bloomberg reported.

In January 2026 Z.AI listed in Hong Kong, and its current market capitalization is about $32 billion, Bloomberg said. On Artificial Analysis’s leaderboard, read on 19 October, GLM-5.3 (max) scored 45 on the Intelligence Index, level with DeepSeek’s newest Flash model, as we reported.

The Vulnerability Ledger

Bloomberg also cited a report by Garnaut Global, a consulting firm run by former national security officials John Garnaut and Matt Pottinger. It said Z.AI’s online ledger, established on 14 August, showed the company found 2,436 vulnerabilities in software from Microsoft, Intel, Google and other Western companies, and disclosed just 5% of them to the companies.

Ten days later Z.AI took the ledger down, replacing it with a notice that the Chinese government would manage vulnerabilities found by its AI models through databases including one run by the Ministry of State Security, the Garnaut report said, according to Bloomberg. Microsoft declined to comment; Intel and Alphabet didn’t respond, Bloomberg said.

The Same Week in Washington

Bloomberg’s report came on 19 October, the same day a Department of Homeland Security order requiring civilian federal agencies to remove DeepSeek products from their IT systems was filed for public inspection, and the State Department sanctioned five entities including former Google engineer Linwei Ding’s startup. China’s Commerce Ministry published export-control announcements covering drone technology the same day.

The Numbers, by Our Arithmetic

Four of the five founders Z.AI lists as “key persons” appear in the project documentation, by our count from Bloomberg’s report. The project’s four years ran from January 2018 to December 2021, overlapping Z.AI’s start in 2019, which Bloomberg noted.

Of the 2,436 vulnerabilities Garnaut Global counted, 5% would be about 122 disclosed to vendors, by our arithmetic.

The Structural Read

As we read it, the overlap Bloomberg describes is in the project documentation: one founder led the project alongside a CGTRI engineer, and three others are listed in its directory of research achievements, according to Bloomberg.

As we read it, the caveats are part of the story: Bloomberg wrote that it’s not clear what, if anything, the founders may have known, and MI5 says some academics may not be aware CGTRI is funding their projects.

As we read it, the report adds to an existing US record on the company: the Bureau of Industry and Security added Zhipu entities to the Entity List on 16 January 2025, and Z.AI said that decision “lacked factual basis,” Bloomberg reported.

Three Implications

A LISTED COMPANY IS INVOLVED Z.AI listed in Hong Kong in January 2026 and has a market capitalization of about $32 billion, Bloomberg said.

MI5 POINTS TO LEGAL ADVICE MI5 says any institution or individual continuing to conduct research ultimately funded by CGTRI should take their own independent legal advice.

THE VULNERABILITY LEDGER IS A SEPARATE THREAD Garnaut Global said Z.AI disclosed just 5% of 2,436 vulnerabilities to the affected companies, according to Bloomberg.

The Business Engineer Lens

This story maps onto the Business Engineer framework AI’s Geopolitical Chokepoint.

The framework puts it this way: “It’s a fundamental constraint on the speed of AI transformation, and it’s creating geopolitical leverage that operates independently of who has the best models.”

As we read it, the Z.AI report sits on the government side of that idea: the actions in our timeline are official US and UK documents, from an Entity List rule to an espionage alert, that act on Chinese AI labs and their research ties regardless of model scores. The framework was published in December 2025 and does not discuss this case.

What Is Not Established

MI5’s alert does not mention Z.AI, Bloomberg wrote. Bloomberg also wrote that it’s not clear what, if anything, the founders may have known about CGTRI’s alleged relationship with Chinese intelligence, and MI5 says some academics may not be aware CGTRI funds their projects.

We have not read the 78-page researchers’ report or the Garnaut Global report ourselves; both are cited here as Bloomberg described them. Z.AI didn’t respond to Bloomberg’s emailed requests for comment, Bloomberg said.

The Bottom Line

Bloomberg reported on 19 October 2026 that four Z.AI founders, including chief executive Zhang Peng, worked on a 2018 to 2021 AI research project co-funded by CGTRI. MI5’s 30 September alert says CGTRI’s primary purpose is to fund research that improves the Ministry of State Security’s technical capability for espionage. The alert does not mention Z.AI, Bloomberg wrote; China’s UK embassy called the allegations against CGTRI baseless, and Z.AI didn’t respond to Bloomberg’s requests for comment.