Sources: the Cyberspace Administration of China is investigating DeepSeek and Moonshot over alleged data leaks to Anthropic via Claude and has questioned staff
THE SO WHAT
China’s probe into DeepSeek and Moonshot over alleged data leaks to Anthropic underscores that model-to-model interactions are now a regulatory and national security concern. If your teams are pasting sensitive artifacts into third-party models, treat that as a cross-border data transfer with real compliance exposure.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIMeta patches Muse exploit that let attackers control the AI agent
Agent shells like Muse are becoming new privilege surfaces—an undocumented setting was enough to turn a local AI helper into a control vector. If you’re rolling your own desktop agents, threat-model them like a browser or password manager, not a toy app.
Applied AIAndy Burnham wants the UK to use its G20 presidency to broker a global AI agreement
If the UK uses its G20 presidency to push a global AI deal, cross-border rules on training data, safety standards, and export controls could harden faster than most roadmaps assume. Multinational teams should start mapping which AI workflows are most exposed to jurisdictional shifts, not just model vendor changes.
Applied AISources: DeepSeek and Moonshot are among the companies briefing the UN Security Council on AI risks and international security during the UN General Assembly
When the UN Security Council is hearing directly from DeepSeek and Moonshot on AI risks, frontier model governance is moving into the same room as hard security policy. Expect more scrutiny of dual-use capabilities and cross-border model access—especially for products that could touch critical infrastructure or information ops.
Applied AIAnthropic and OpenAI ask Australia to ease its ban on AI training with local content
Anthropic and OpenAI pushing Australia to relax its training-content ban shows that model quality is now constrained by national copyright regimes. If you rely on local-language or domain-specific performance, track where your vendors can legally train—capability gaps may start to mirror regulatory borders.