Kinling Lo reports on open-source AI in China and the U.S. for Rest of World

Kinling Lo reports on open-source AI in China and the U.S. for Rest of World

Rest of World published Kinling Lo’s interview with Tiezhen Wang on June 15, 2026. Wang argues that open-source AI can make the China-U.S. competition collaborative: he says Chinese labs’ releases help U.S. labs, and that both sides can benefit from growing the field together. He describes monetization as the challenge: labs may earn through infrastructure, paid services, or branding, but cloud providers can profit from running a model without sharing revenue. Wang considers payment requirements for commercial use fair, while warning that labs’ inability to fund research could push them away from open-source releases. Wang also argues that model distillation is ordinary research practice, and says AI-generated content should have zero copyright. He describes a possible startup path from closed-source models toward open-source models as a way to reduce costs, and says Chinese companies’ access to less expensive open-source models could encourage faster adoption. He reports that some companies are giving employees unlimited tokens and says he would not be surprised by substantial AI use in China over the next year or two. These are Wang’s assessments and forecasts; the interview does not establish how broadly the practices he describes apply or whether his predictions will occur.

Publication: 4dfde542-e1ab-6037-b077-14a2e37f9ff1

attribution — pass: This is a single interview article by Kinling Lo for Rest of World. The reading attributes the arguments, examples and forecasts to Tiezhen Wang, and it correctly identifies Lo as the interviewer and author. The title's 'reports on' is broad but not misleading.

By Kinling Lo 15 June 2026

summary — uncertain: The summary follows Wang's main arguments faithfully, but two central terms are left unexplained for a general reader. 'Tokens' is never explained, so 'giving employees unlimited tokens' and the cost argument are hard to follow. Smallest repair: explain that tokens are the metered units of AI usage that companies pay for. 'Open-source' is also not glossed, although the source explains it as models developers can download, inspect and customize for free. Only distillation is defined beside the reading. Wang's force on distillation, including his argument about crawling the internet, is somewhat flattened but still present.

giving their employees unlimited tokens just to see what they can do

date — pass: evidence.originalPublished (2026-06-15 UTC) matches the source dateline of 15 June 2026. Wang's forecast is reported as his attributed expectation, not as an outcome. There is no hindsight framing. This is a publisher page, so no archive disclosure is needed. The dateModified is nine minutes after publication, and editedAfterPublication is not flagged.

By Kinling Lo 15 June 2026

support — pass: The claims are grounded in Wang's statements. Listing 'branding' as a way to earn is a slight stretch: the source presents branding as a reason to open-source, mainly to attract talent, though it appears within the monetization answer. Concerns about funding and closed-source drift, revenue-sharing with cloud providers, and cheaper models enabling faster adoption are all supported and attributed to Wang. The limitations correctly note that these are his assessments.

If they do not find a way to monetize their research, this is a real risk.

media — pass: No media was selected, and no verified media candidates were supplied, so nothing needs checking.