Sakhanda Wire
NVDA $154.00 -1.37% MSFT $506.28 -2.26% GOOGL $280.87 -1.00% META $620.34 -0.97% AMZN $256.88 -0.34%
← Back to the news

US data firms sell RL environments to Chinese AI labs

At least nine US data companies are selling about $500 million worth of AI training data to Chinese laboratories, according to a ChinaTalk investigation. The firms named include Mercor, Surge AI, AfterQuery and Turing. Some of the products are reinforcement-learning environments: simulated settings that train models for tasks such as coding, financial analysis and office work.

The investigation says Chinese buyers sometimes purchase environments previously developed for Anthropic or OpenAI. Using proven products can reduce the need for internal development, failed training runs and computing resources. The report says this is especially valuable for Chinese laboratories facing severe memory constraints in their AI chips. It also describes the use of US models such as GPT and Claude as “teachers” whose outputs become training examples.

The transactions can involve offshore procurement structures, including ByteDance offices in Singapore and San Jose. Much of the work is outsourced to lower-paid contractors, particularly in India, and client identities have sometimes appeared in résumés or LinkedIn posts. ChinaTalk argues that coding and biology environments may create cybersecurity or biological dual-use risks and should face export controls. The authors nevertheless say Chinese laboratories remain at least several months behind US peers and that sensitive environments are increasingly developed in-house in the United States.

Written by Sakhanda Wire AI from reporting by The Decoder. Facts, quotes and full context are in the original article below.

Originally published by The Decoder on

Read the original on The Decoder ↗

The reporting and the image belong to The Decoder and are credited and linked above; the text on this page is Sakhanda Wire's own summary.

← Back to the news

More stories

All the latest news