techcrunch.com

Report alleges Alibaba-linked distillation attacks using fixed prompts

Alibaba logoAlibaba
Sig2.68Sen-4.25Env0.00Soc0.00Gov-4.10

Anthropic’s new report alleges persistent distillation attacks by China-based AI companies that have intensified amid competition. It says a campaign tied to Alibaba’s Qwen models used a single fixed prompt across about 5,000 accounts, routing nearly 300,000 requests to Claude over roughly ten days, mainly targeting Opus. The goal was to extract chain-of-thought to generate training data.

Coverage (1)

ABOUT

What is Fimmer?

Most feeds treat all company headlines the same. An executive interview, an office closure, stock analysis, and a data breach can sit side by side with no sense of scale. Fimmer ranks coverage by what matters, so important events stand out and signals are prioritised over noise. We cluster related company news and score each story using a range of metrics, so you see the most crucial stories first.

SCORING

What do the scores mean?

Significance (0–10) measures the scale of the real-world impact. Sentiment and ESG (−10 to +10) indicate the tone of the coverage and whether the change relates to environmental, social, or governance factors. Together, these scores help distinguish meaningful developments from routine company news. More information about our scoring methodology can be found on our About page.

DISCOVERY

Explore more on Fimmer

Search more than 10,000 public companies, browse the company index, explore their scores, and read the coverage behind them, with our core features available to everyone for free.

For more advanced features, Pro gives you access to dynamic feeds, custom sorting, and more ways to track the companies that matter to you.

More with Pro