Trump Bans Anthropic from US Government; Pentagon Picks OpenAI
President Trump ordered all federal agencies to cease using Anthropic’s AI products, and Defense Secretary Pete Hegseth designated the company a “supply chain risk to national security.” Hours later, OpenAI CEO Sam Altman announced a deal to supply AI to classified Pentagon military networks — a dramatic realignment of the AI-military relationship.
This follows Anthropic’s refusal to cooperate with certain military use cases, putting it at odds with the current administration’s approach to AI in defense. The move effectively makes OpenAI the primary AI vendor for the US government’s most sensitive operations.
🔗 NPR | NBC News | NYT | The Guardian
DeepSeek V4 Expected This Week — Multimodal, Huawei-Optimized
Chinese AI startup DeepSeek is poised to release its next-generation V4 model as early as March 3. V4 is multimodal — generating images, video, and text — and has been optimized for Huawei and Cambricon AI chips, reducing dependence on Nvidia hardware.
The timing is strategic: landing just before China’s NPC session and while US-China AI tensions are at fever pitch. Wall Street is bracing for a potential repeat of the January 2025 “DeepSeek shock” that wiped billions from tech valuations.
🔗 PYMNTS | Futurism | Evolink
China’s NPC Opens Thursday — 15th Five-Year Plan Sets AI & Tech Roadmap
China’s National People’s Congress begins Thursday with the release of the 15th Five-Year Plan (2026–2030). The blueprint is expected to prioritize AI, humanoid robotics, supply chain resilience, and advanced semiconductors. The GDP growth target is anticipated at 4.5–5.0%.
Last year’s plan mentioned AI models for the first time; this year, “embodied intelligence” (humanoid robots) and domestic chip supply chains are expected to receive major policy support and funding.
🔗 Reuters | Bloomberg
Anthropic Accuses Chinese AI Labs of “Industrial-Scale” Distillation
Anthropic publicly accused three Chinese AI companies — MiniMax, Moonshot, and DeepSeek — of coordinated campaigns to extract chain-of-thought training data from Claude. MiniMax alone allegedly conducted over 13 million exchanges. Google and OpenAI have filed similar complaints.
Anthropic warned that distilled models may lack safety guardrails, creating potential national security risks. OpenAI’s memo to the House Select Committee on China called it “ongoing efforts to free-ride” on US-developed technology.
🔗 CNBC | NYT | CNN
Sam Altman on AGI Alignment: “Humanity Must Stay in Control”
In a late-night thread on X, Sam Altman pushed back on colleagues who argue that a benevolent AGI losing human control would be acceptable: “I personally think it’s really important humanity stays in control.” He called long-term alignment “one of the most important areas” of research — notably timed as OpenAI deepens military ties.
🔗 X/Twitter
MWC 2026: Honor’s Humanoid Robot, Robot Phone, and More
Mobile World Congress 2026 kicked off in Barcelona with a flood of announcements:
- Honor debuted a dancing humanoid robot and a morphing “Robot Phone” concept, backed by a $10B five-year AI commitment
- Xiaomi 17 Ultra launched globally with professional camera rig
- Lenovo Legion Go Fold — a foldable gaming handheld concept with 11.6” flexible display
- Qualcomm Snapdragon Wear Elite — new chip targeting AI-powered wearables
- Motorola Razr Fold — flagship foldable with silicon-carbon battery
- Apple expected to hold an “Apple Experience” event Wednesday
🔗 Engadget | The Verge | CNET
“Oh Great, Here Comes 6G” — Satellite and AI-Native Networks at MWC
The Verge covers the 6G push at MWC Barcelona, focusing on Integrated Sensing and Communication (ISAC), LEO satellite convergence with terrestrial networks, and AI-native network architectures. Commercial 6G products remain years away, but the standards race is accelerating.
🔗 The Verge
OpenClaw 2026.3.1 Released
The latest OpenClaw release brings OpenAI WebSocket streaming, Claude 4.6 adaptive thinking support, improved Docker and native Kubernetes support, Discord thread handling, and Feishu fixes. The team also introduced an agent-powered visual diffs plugin.
🔗 X/Twitter
Markets Brace for Volatile Week
Geopolitical tensions and surging oil prices triggered a sharp defensive pivot in premarket trading Monday. AI sector valuations face renewed scrutiny as investors weigh the Anthropic government ban, imminent DeepSeek V4 release, and China’s NPC policy signals.
🔗 Reuters
Cory Doctorow: “No One Wants to Read Your AI Slop”
The latest Pluralistic essay tackles the growing backlash against AI-generated content flooding the internet. Doctorow argues that “nonconsensual slopping” — being forced to consume AI-generated content you didn’t ask for — is becoming the defining user experience problem of 2026.
🔗 Pluralistic
Takeaway
The US-China AI cold war just went kinetic: government bans, industrial-scale IP theft accusations, DeepSeek V4 about to land, and China unveiling its 5-year tech blueprint — all in the same week. The battle lines between open vs. closed, East vs. West, and military vs. civilian AI are being drawn in real time.