Implications For Your AI in Business – Did Chinese AI Companies Copy ChatGPT, Claude and Gemini?
One thing that I appreciate about the current time is at least with the US, there is little ambiguity. They’re broadcasting their intentions out loud, but for years I’ve already understood the silent message and how to protect myself but does your company?
“China-based AI companies that conduct industrial-scale distillation against US AI models see significantly shorter AI development timelines and reduced financial expenditures in training a frontier model,” agencies said. — https://arstechnica.com/tech-policy/2026/09/six-chinese-ai-firms-accused-of-aggressively-copying-us-frontier-models/
I recently wrote about how the US is getting private companies to hack “adversaries” and my skepticism about what it was really about. And now today the US releases a report claiming that all of the Chinese AI companies have created nothing and are a complete copy of US models through exfiltrated data. This is hardly of relevance to what I’m going to be discussing but I’ll address it quickly.
Let me just ask this one simple question. If you robbed a bank would you be sharing the money in full view with pictures and blueprints of how you robbed the bank? That is what the US expects us to believe, that in particular, Alibaba’s Qwen and DeepSeek would copy US AI, arguably improve it, and then release the full open source, copied models. The only better one from the US lately is what I describe as “The Worst Chinese Hackers” because it appears breadcrumbs leading straight back to China were intentionally and conveniently left. But regardless of what any of us believe, this point is actually irrelevant because the US response to its allegations are all that matter at this point.
What Will The US Do?
I already mentioned they have now given the greenlight for US IT companies to openly conduct cyberattacks.
AI from the US will now become more locked down and risky to use because of the above and the US is not shying away from this:
All American AI firms must work with the government and US allies to end the alleged theft threatening the US lead in the AI race, the agencies said. That will require coordinated action across the AI ecosystem to combat the “aggressive, malicious, and targeted distillation activities at an industrial scale that extract restricted proprietary functionalities and capabilities of US frontier AI models.” — https://arstechnica.com/tech-policy/2026/09/six-chinese-ai-firms-accused-of-aggressively-copying-us-frontier-models/
The next phase is what I will describe as the precursor before malicious responses as their guidelines imply:
Next, agencies asked firms to start dumbing down model responses when suspected distillation attacks are flagged. By “subtly” altering responses—such as by “presenting correct information with different reasoning,” adding stylistic inconsistencies, or reducing reasoning depth—firms can decrease the payoff for Chinese firms. US firms could also secretly switch malicious accounts to an inferior model, and they should do so without providing any notice, agencies suggested.
Obviously the issue above is “what makes you a bad Chinese user?” US based AI companies now have arbitrary permission not to only give you harmful or incorrect responses at worst, or at best they have downgraded the accuracy and model that is being used, all quietly and secretly of course.
I’ve already talked about the dangers of US AI, the Huggingface incident and how I do not believe that is an accident or “out of control AI”; and how Anthropic has been caught injecting hidden code in its Claude Agentic AI service.
Today bad users to the US can be defined as “Chinese” but is it really going to end with China? The US considers Canada an adversary and has openly threatened EU countries.
I’ve said it repeatedly that your only chance of being safe is to use the Chinese, OpenSource models (preferably self-deployed like Qwen 3.8 27B so you can actually see what they are doing). Whether you are from Canada, the US, EU or anywhere else in the world, the argument today is even stronger that the use of any US AI services now carry increasingly unacceptable risks.
What do you think, would you still trust US AI given the recent developments and does your company understand the implications or take security seriously? If you need help with your AI Strategy and Cybersecurity from experts that actually understand IT and AI, then contact myself or Techrich.
