A fight over how AI models should be released has broken into the open, and the usual battle lines are not where you might expect.
A Yahoo Finance report, carried via Google News, asks why Nvidia, Meta and Microsoft are publicly defending open-weight models like Kimi K3. "Open-weight" refers to models whose trained parameters — the weights — are released for anyone to download, run on their own hardware and adapt, rather than being locked behind a company's paid interface.
According to a Techmeme summary of a piece by M.G. Siegler at Spyglass, US tech giants have shifted their stance toward publicly backing open AI models, with Anthropic and Amazon remaining notable holdouts, alongside the US government. Siegler frames the moment around an open letter on open weights, which he describes as feeling familiar yet different, and draws a comparison to Steve Jobs' famous "Thoughts on Flash" open letter.
The sources available here do not spell out each company's reasoning, so it is worth being careful about motives. What they do establish is the shape of the disagreement: several of the largest US technology companies now want to be seen arguing for openness, while at least two prominent players — one of them a leading model developer, the other the largest cloud provider — have not joined them.
Why it matters: whether the most capable AI models are freely downloadable or tightly controlled determines who gets to build on them, who can inspect them, and which handful of companies sit at the center of the industry.