Given all that North Korea has shown itself to be capable of exfiltrating and hacking, and all that many private hackers have been able to steal from multiple major companies with "good security", I have little doubt that a hugely powerful and capable state like China's could indeed break into OpenAI's data if they really wanted to. Though, if they could, I imagine they'd be subtle about it, maybe even having employees in this or other companies working there as deep cover agents, possibly for years. No, none of this is absurd. The USSR did it effectively for decades even digging into undertakings as deeply sensitive as the Manhattan Project.