The companies you listed in contrast to OpenAI also have some key differences: they're all long-standing and mature companies that have been through several management and regime changes at this point, while OpenAI is still in startup territory and hasn't fully established what it will be going forward.
The other major difference is that OpenAI is split between a non-profit and a for-profit entity, with the non-profit entity owning a controlling share of the for-profit. That's an unusual corporate structure, and the only public-facing example I can think of that matches it is Mozilla (which has its own issues you wouldn't necessarily see in a pure for-profit corporation). So that means on top of the usual failure modes of a for-profit enterprise that could lead to the CEO getting fired, you also get other possible failure modes including ones grounded in pure ideology since the success or failure of a non-profit is judged on how well it accomplishes its stated mission rather than its profitability, which is uh well, it's a bit more tenuous.
In wartime, pandemics, and in matters of national security, the government's power is at its apex, but pretty much all of that has to withstand legal challenge. Even National Security Letters have their limits: they're an information gathering tool, the US Government can't use them to restructure a company and the structure of a company is not a factor in its ability to comply with the demands of an NSL.