The US is the dominant player in AI, software, robotics etc. Its need for chips will not recede relative to other regions in the coming quarters, if anything it may increase.
Unlike 5 years ago, today practically every person with power or influence (Military, Private Sector, Finance, Academia etc.) understands the value of TSMC's chips & will pay a premium to stockpile chips as an insurance policy against a possible invasion of Taiwan.
LLMs are stochastic & effective precisely because the delta between human responses & LLM responses is so small. Therefore, such labeling is bound to end up with a number of false-positives & not particularly helpful for users. The ROI on implementing such a feature, especially before June, renders it highly unlikely.
However, AFAIK OpenAI isn't using these techniques just yet, which makes it very unlikely that social media are going to be able to label text as potentially written by AI