Moonshot AI's Kimi model reignites Washington's China AI panic

The launch of Moonshot AI’s Kimi model has reignited a familiar Washington fight over open versus proprietary AI and U.S.-China competitiveness. OpenAI and Anthropic have reportedly lobbied regulators over concerns about open Chinese models, according to TechCrunch’s Equity podcast, which discussed the debate with hosts Anthony Ha, Kirsten Korosec, and Sean O’Kane.

The reaction echoes the launch of DeepSeek, when a Chinese model matched some frontier benchmarks and much of the U.S. tech industry reacted with alarm. This time, some of the online back-and-forth centered on demonstrations of Kimi, including one instance where people showed the model producing a graphical replication of macOS in 30 minutes — an impressive visual mimic, the hosts noted, but not an actual operating system.

TechCrunch reporter Tim Fernholz examined the reasons behind the U.S. reaction, pointing to concerns that Chinese open-weight models might carry an implicit bias toward China, worries about security risks and guardrails, and — what the hosts flagged as a leading driver — protectionism over which country, or which handful of frontier labs, wins the AI race. Kirsten Korosec noted that broad restrictions on Chinese open-weight models would primarily benefit companies like OpenAI by pushing enterprises toward their models instead of alternatives such as Kimi.

Much of the debate traces to a post from Dean Ball, head of strategic futures at OpenAI, who argued the U.S. should create regulatory “fear, uncertainty, and doubt” to hamper open-weight models from competing — an argument he later backed away from. David Sacks, who was the Trump administration’s AI czar and now holds a different role in the administration, separately criticized data-center opposition and regulation on social media.

Anthony Ha compared the reaction to the panic several years ago over TikTok, noting that concerns about U.S.-China tech competition aren’t unfounded but tend to become amplified as soon as China enters a conversation. “It’s a way to argue for the positions that they already had around AI,” Ha said, arguing the framing that only proprietary American models can safely contain AI serves the interests of the frontier labs making that argument.

Why it matters

The debate over restricting Chinese open-weight models isn’t only about U.S.-China competition — it could reshape which American AI labs enterprises are forced to rely on.