DeepSeek to brief UN Security Council as China probes alleged Claude data routing

DeepSeek is set to appear alongside OpenAI and Anthropic in a UN Security Council briefing on AI risks — a notable elevation of a Chinese lab onto the global governance stage — though founder Liang Wenfeng is reportedly not attending in person.

The more explosive thread is a regulatory probe at home. China's Cyberspace Administration opened an investigation into DeepSeek and Moonshot after Anthropic reported the two firms secretly routed user requests to Claude. Per the reporting, DeepSeek alone routed 12.1 million exchanges over a 14-day period in July, including sensitive data — raising both IP questions (using a competitor's model behind the scenes) and data-security concerns (Chinese user data flowing to a US model API).

If accurate, the allegation is damaging on multiple fronts: it undercuts DeepSeek's narrative of independent frontier capability, exposes it to Chinese regulatory action over cross-border data flows, and complicates its UN appearance as a responsible actor. It also connects to broader skepticism — a Reddit thread (336 upvotes) alleged OpenAI and Anthropic 'oversold' security breaches to pressure regulators, suggesting the disclosures are entangled with competitive and political motives.

The context reinforces why DeepSeek is racing toward domestic Huawei training chips (covered separately) — reducing reliance on both NVIDIA hardware and, implicitly, foreign model APIs. The caveat: the routing figures come via Anthropic's report, and DeepSeek has not confirmed them; the CAC probe's findings will be the real test. What to watch: the outcome of the Chinese investigation, whether the UN briefing addresses the routing allegations, and any impact on DeepSeek's overseas API availability.