Archived (removepaywall.com)

Four founders of one of China’s most promising artificial intelligence companies, Z. ai [also known in China as Zhipu AI], worked on research in a joint project with a technology institute that the British government accused of having “very strong ties” to China’s main spy agency, according to publicly available research papers from China and an “espionage alert” from Britain.

The nearly four-year AI research project, which overlapped with the company’s start in 2019, was co-funded by Beijing-based China General Technology Research Institute (CGTRI) and the National Natural Science Foundation of China, according to a previously unreported 78-page report that the project’s leaders compiled to detail their findings.

A CGTRI engineer and one of Z ai’s founders were among the project’s leaders, and three other Z. ai founders contributed to the research, according to the report.

[...]

An additional report says Z. AI co-founder Li Juanzi, a professor at Tsinghua University, led the research alongside CGTRI engineer Cao Weibing and a Southeast University professor. Three additional Z. AI founders contributed to the work:

  • Tang Jie, listed as co-author on project papers
  • Xu Bin, named as co-inventor on related patents
  • Zhang Peng (current Z. AI CEO), holder of software copyrights from the project

[...]

Separate analysis by national security consulting firm Garnaut Global revealed troubling patterns in Z. AI's handling of software vulnerabilities discovered by its AI models: According to the firm's report, Z. AI established a public ledger on August 14, 2026 documenting 2,436 vulnerabilities found in software from Western technology companies, but disclosed only 5% to affected companies before routing data to Chinese government.

Rather than following standard responsible disclosure practices, Z. AI disclosed only 5% of these vulnerabilities to affected vendors. Ten days after posting the ledger, the company replaced it with a notice stating the Chinese government would manage vulnerabilities through databases including one operated by the Ministry of State Security.

James Mulvenon, who leads intelligence research at national security contractor Pamir Consulting, characterized the situation as giving China's peak cyberespionage agency access to exploitable weaknesses in critical Western software systems.

[...]

all 19 comments

sorted by: hot top controversial new old
[–] 27 points 3 weeks ago

Anthropic have engineers stationed in NSA offices

  • source
  • [–] 12 points 3 weeks ago*

    I thought it should be fairly obvious to people not to rely on code run from Chinese servers in general for security or privacy.

    Also, not any worse than the US based services, who already work with US government directly and via secret courts and backdoors -- they also do not report vulnerabilities, and in the case of Claude, will allow you to find zero of them.

    In general though, you should be running these on your own servers or compute you control or can trust.

  • source
  • [–] 32 points 3 weeks ago (8 children)

    Same as the US. Somehow, I'd rather have the Chinese steal my data and be able to do nothing meaningful with it, than have the US steal it and subvert my democracy.

  • source
  • hideshow 8 child comments
  • [–] 3 points 3 weeks ago

    I think about that sometimes as well. Less about subverting western democracies, because multiple parties foreign and domestic have that goal. But more about which government actually has authority over me. My own government does. And they cannot be trusted.

  • source
  • parent
  • [–] [S] 6 points 3 weeks ago (5 children)

    China isn't subverting democracy?

    Chinese steal my data and be able to do nothing meaningful with it.

    Is this satire?

  • source
  • parent
  • hideshow 5 child comments
  • [–] 10 points 3 weeks ago (4 children)

    Which Western country's democracy is being subverted by China with my data, and how? I genuinely want to know.

    I know Facebook, Instagram, YouTube, Twitter, Reddit, and others target citizens and radicalise them to destabilise countries. Their private data is also sold to "think tanks" who work with politicians to influence people to vote in a specific way. How is China influencing Western democracies. Tell me. Link to articles so I can read up on it.

  • source
  • parent
  • hideshow 4 child comments
  • [–] 7 points 3 weeks ago (1 child)

    Their ZCODE harness quite literally got caught zipping up people's entire git trees and uploading them to alibaba servers. I have no qualms with Chinese ai but this specific company is on my blacklist.

  • source
  • hideshow 1 child comment
  • [–] 6 points 3 weeks ago

    Just like facebook!

    . . . err

  • source
  • [–] 13 points 3 weeks ago

    So basically the same as American AI companies?

    Except only one of these two may actually benefit anyone other than billionaire pedophiles, while the other contributes to genocide.

  • source
  • [–] 4 points 3 weeks ago
    [–] 9 points 3 weeks ago*

    Check how much technological research in USA is funded by the Dept of Defense... Basically all of it.

    https://www.executivegov.com/articles/pentagon-funding-autonomous-weapons-hegseth-50-billion

  • source
  • [–] 3 points 3 weeks ago* (last edited 3 weeks ago)

    I mean ofcourse they do, and so does probably some/all western AI companies. People share all their secrets with these tools, I know people who were previously obsessed with being ghosts in the internet now running AI sessions in their computer (non local) with full access. This level of reach is a wet dream for intelligence agencies.

  • source
  • [–] 2 points 2 weeks ago

    So it's a safe investment?

  • source
  • 4 comments in c/china@sopuli.xyz view that post
    [–] 4 points 3 weeks ago

    Isnt there a law in China that disclosures must be sent to their government first before being sent to the software vendors? So they would be breaking the law if they didnt do that?

    Also:

    Zhipu AI (Z.ai)

    We build the ChatGLM family of LLMs, develop LLMs as Agents, and release related LLM training & inference techniques:

    GLM-5.3 & GLM-5.2 & GLM-5.1 & GLM-5, GLM-4.7 & GLM-4.6 & GLM-4.5, GLM-V, GLM-OCR, GLM-ASR, GLM-TTS, GLM-Image, CodeGeeX, CogVLM (VisualGLM), WebGLM, GLM-130B, CogView, CogVideo && CogVideoX. CogAgent, AutoWebGLM, AgentTuning, APAR.

    We also work on LLM evaluations: LVBench, MotionBench, ComplexFuncBench, AgentBench, AlignBench, LongBench, NaturalCodeBench.

    We started with social networks and graphs, and always love them: GraphMAE, GPT-GNN, GCC, SelfKG, CogDL. AMiner.

  • source
  • [–] 2 points 3 weeks ago (2 children)

    Like western frontier model companies!?

  • source
  • hideshow 2 child comments