Moonshot AI reports false arrest rumors to police after K3 launch

In July, Moonshot AI released the largest free artificial intelligence model globally. Eight weeks on, it is filing police complaints.

The Kimi K3 model appeared on July 16, freely downloadable and runnable by anyone. Such open access is unusual for an industry that typically keeps top models private. Then the systems began to fail under strain.

Launch Overwhelms Its Own Infrastructure

Four days into the launch, Moonshot stopped selling fresh subscriptions. Paying members retained their seats while others were placed in a queue.

“Over the past 48 hours, demand has pushed close to the limits of our current capacity,” the team said in a statment.

That proved to be the company's most prominent public statement over the summer.

On September 11, Moonshot deployed a smaller model, K2.8 Preview, on its code platform. Scanning the changelog closely shows something unusual. Some K3 queries are now served by the smaller model instead.

🚨 Kimi K2.8 Preview just dropped in Kimi Code

From reading the docs it has performance close to K3 but apparently with "more efficient thinking"

• 1M context
• image + video input
• low / high / max reasoning

Kimi K3 is a very good model too, im hoping the "more efficient… pic.twitter.com/52I5Ege2TV

— Lumina (@LuminaBench) September 11, 2026

That same week, however, Anthropic acused Moonshot of quietly forwarding Kimi users' inquiries to Claude. It claimed Moonshot displayed Claude's responses as Kimi's own work. Within a single 10-day span, it registered roughly 300,000 transfers routed via 5,380 fake accounts.

Anthropic says one such user was likely military. They uploaded footage from hundreds of security cameras in Chengdu and asked whether a man was acting strangely.

The awkward part is that all of this relies solely on Anthropic's own logs. No regulator has validated it. BeInCrypto reported on September 1 that Anthropic had already deployed a block against this. That occurred ten days before its September threat report named any firm.

After that, a rumor spread that Moonshot's founder had been taken in after the Anthropic report.

网传Kimi被带走16个人，包括老大，如果传闻属实，那大概率可能也是和Anthropic刚发布的入侵报告有关。在最后一节“非法蒸馏里”，Anthropic说月之暗面把本该发给 Kimi 处理的客户请求偷偷转给 Claude，再把 Claude 的回答当作 Kimi…

— ZhangChi (@zhangchitc) September 11, 2026

On Saturday, the company answered.

“The information circulating online regarding the founder and employees is completely fabricated and malicious slander. We have immediately reported the matter to the police and will pursue legal action against those responsible for spreading the rumors,” Moonshot AI said on Saturday.

No newspaper has confirmed an arrest. No agency has confirmed an investigation. What is confirmed is odder. The year's largest open AI model is now partly handled by a smaller one, and the fullest account of where its users' queries went was written in San Francisco.