BABA
Another AI drama, now with subpoenas-adjacent vibes
Alibaba is back in the market’s crosshairs after Anthropic accused the Chinese giant and its Qwen AI unit of orchestrating a large-scale campaign to extract know-how from Claude. In plain English: Anthropic says someone on Alibaba’s side tried to teach a smaller model by vacuuming up answers from a smarter one. Very “copy my homework, but don’t make it obvious.”
The allegation in a nutshell
According to a letter Reuters reviewed, Anthropic told U.S. lawmakers that operators affiliated with Alibaba used nearly 25,000 fraudulent accounts to generate more than 28.8 million Claude exchanges between April 22nd and June 5th.
Anthropic says the goal was to help China build AI systems that could catch up to its Mythos Preview models. Alibaba hasn’t publicly responded yet, which is usually not the kind of silence investors love.
Why the market cares
This is the sort of headline that can spook investors in two directions at once:
- it raises legal and regulatory risk around Alibaba’s AI push
- it adds a fresh trust problem just as the company is trying to look like a serious cloud-and-AI player, not just an e-commerce behemoth with a side hustle
Shares were down about 2.79% at the time of publication, which is the market’s way of saying: “cool story, please fix the optics.”
Big picture: when an AI company gets accused of harvesting another model’s output at scale, that’s not just a PR headache. It can snowball into tougher scrutiny, more geopolitical friction, and a cloud over the whole AI growth story.