OpenAI and Anthropic Line Up Against Open Weights

Axios reports that OpenAI and Anthropic — usually happy to disagree in public — have converged on the same message in Washington: open-weight models, and Chinese ones in particular, are a security problem the Trump administration should act on.

The obvious counter, and the one that dominated the 330-comment Hacker News thread, is that freely downloadable frontier-ish weights are also the single most direct threat to per-token API revenue. Axios's own headline makes the point: risks to their bottom line.

We're keeping this short because the reporting is thin on specifics — no bill text, no named policy ask. Treat it as positioning, not policy.

Sources: Axios, HN discussion