News Watchmen

The Brief

The artificial-intelligence race between the United States and China just entered another contentious phase.

Anthropic says Chinese AI companies carried out massive campaigns to extract capabilities from Claude and use them to improve competing models—including more than 151 million exchanges attributed to Alibaba alone.

In its latest threat-intelligence report, Anthropic said it identified and disrupted illicit distillation campaigns involving several China-based labs, including Alibaba, Moonshot AI, DeepSeek, Zhipu and Xiaomi. The company says the activity involved nearly 200 million exchanges in total.

Anthropic says Alibaba’s campaign was the largest it has ever measured.

According to the company, Alibaba operated more than 3,500 fraudulent accounts and generated nearly 3 million Claude exchanges per day at its peak between May and July. Anthropic says the harvested reasoning transcripts were used to help train Alibaba’s Qwen models.

How The Operation Allegedly Worked

The technique is known as knowledge distillation.

Distillation itself is not inherently illegal or unusual. AI developers commonly use a more capable “teacher” model to generate outputs that help train a smaller “student” model.

The controversy is over how the data is obtained.

Anthropic says the companies used proxy services, fraudulent accounts, stolen credentials and other methods to access Claude at scale despite geographic restrictions and terms of service.

The company says Moonshot routed more than 23 million exchanges through Claude between May and July, while DeepSeek accounted for more than 12.1 million exchanges over a 14-day period in July.

Anthropic also says some of the rerouted conversations contained sensitive corporate, government and military-related information that users may not have known was being sent to Claude.

Washington Is Watching

This is no longer merely a corporate dispute.

On September 8, the FBI, NSA and CISA accused six Chinese AI companies—including Alibaba, DeepSeek and Moonshot—of conducting industrial-scale distillation campaigns against U.S. frontier AI models.

China has rejected the allegations, calling them unfounded and arguing that distillation is a legitimate technological technique.

Treasury Secretary Scott Bessent has previously warned that sanctions and Entity List designations could be considered if Chinese companies cross the line from legitimate distillation into intellectual-property theft.

Why This Matters

The stakes go far beyond who produces the best chatbot.

If advanced AI capabilities can be copied or extracted at enormous scale, the competitive advantage of developing those systems could shrink dramatically.

And there is a national-security dimension.

U.S. officials have warned that advanced AI capabilities can support cyber operations, intelligence collection and military applications.

At the same time, the controversy demonstrates another emerging problem:

AI systems may be learning from one another in ways their users—and even their customers—don’t realize.

The Prophetic Connection

Scripture repeatedly warns that knowledge and human capability can increase dramatically while human wisdom does not necessarily keep pace.

Daniel 12:4 speaks of a time when “knowledge shall be increased.”

Modern artificial intelligence is not itself a fulfillment of that prophecy.

But the extraordinary acceleration of technological knowledge should remind Christians of something important:

Humanity’s ability to build powerful tools does not automatically give humanity the wisdom to control them.

The question isn’t simply how intelligent our machines become.

It is whether human beings remain wise enough to use that intelligence responsibly.

What Should We Watch?

Watch for three developments:

1. U.S. sanctions: Will Washington actually place any Chinese AI companies on restricted lists?

2. Trump-Xi talks: AI is emerging as another major issue in the U.S.-China relationship ahead of the planned September summit.

3. The AI arms race: As models become more capable, expect increasing battles over chips, data, intellectual property and access to frontier systems.

The race for AI dominance is becoming a race over knowledge itself.

Prepare For Uncertain Times

The technology landscape is changing faster than most people realize.

Businesses and individuals should understand what data they are giving AI systems, where that information may travel, and what privacy protections actually apply.

Related News Watchmen Coverage

• Trump Administration Blocks New Chinese Humanoid Robots as Washington Warns of Surveillance, Cyberattack and Remote-Control Risks
• Apps Marketed to U.S. Troops Found Carrying Chinese and Russian Code, Raising New National Security Concerns
• Nuclear Material “That Could Be Weaponized” Discovered In Syria — IAEA Raises Alarm
• IIsrael Prepares For Possible Unilateral Strike On Iran As Tensions With Washington Escalate

Frequently Asked Questions

1. What is AI distillation?
It is a training technique in which a smaller AI model learns from the outputs of a more capable model.

2. What does Anthropic accuse Alibaba of doing?
Anthropic says Alibaba used thousands of fraudulent accounts and proxy services to obtain Claude outputs at enormous scale and use them to improve Qwen models.

3. How many exchanges were attributed to Alibaba?
Anthropic says it observed more than 151 million exchanges attributable to Alibaba between May and July 2026.

4. Did China admit to the allegations?
No. Chinese authorities have rejected the U.S. allegations and described distillation as a legitimate technological technique.

5. Could the United States impose sanctions?
U.S. officials have said sanctions and Entity List designations could be considered if alleged distillation crosses into intellectual-property theft.

---

Affiliate Disclosure:
Some links in my articles may bring me a small commission at no extra cost to you. Thank you for your support of my work here!