Policy
Regulation and governance
-
Policy
The Wall Around the Warehouse: Trump's Compute-Shoring Threat
"We are going to keep it all in this country." The threat landed at a press briefing, with the cabinet hearing it for the first time in the room, and it puts a fence around every overseas data center the American giants have promised.
-
Policy
879,000 Chips Through the Front Door
The House release says BIS approved 879,000 of the 3.4 million chips Nvidia and AMD applied to ship to China, and the buyers included two companies on the military list, and every application said the diversion risk was low.
-
Policy
The Last Door Closes: Jensen's Exit Leaves China Two Numbers
Nvidia will no longer sell the H200 to China, and Washington has barred the chips from the data centers that serve China remotely. What is left is a race between two numbers, 34 percent and 28 percent, and a gap of six points that is the whole war.
-
Policy
The Reaffirmation Tour: How Asia Answered Jensen Huang's Exit
Seoul confirmed the order book, Beijing said "not aware," Singapore held the passport, and the chip channel kept running. Day five of the saga, mapped government by government.
-
Policy
The Eye Shifts: Jensen Left the Room, and Huawei's Flagship Rises
Le Monde has named Huawei the conductor of China's self-sufficiency. Now that the man who kept the chip door open is gone, the story the world follows is no longer the door. It is the flagship.
-
Policy
The Peak-Exit Pattern: What Jack Ma's Disappearance Teaches About Jensen Huang's Resignation
An AgentBear exclusive. Two of the most powerful founders on the planet walked away at the top of their power. Neither exit was the story. The mechanism is.
-
Policy
The Weights Remember: Blackwell Fingerprints Inside China's Top Models
A former OpenAI engineer searched over six trillion chip-and-seed combinations and found banned Nvidia silicon in the DNA of DeepSeek's V4-Pro, Kimi's K3, and MiniMax's M3.
-
Policy
The Woman Who Still Won't Show Up: Alice Huang, After the Resignation
Jensen Huang is gone, a third affair has surfaced, and the Singapore shell-company trail behind the chip channel remains the one thread nobody can pull.
-
Policy
Bioweapons Are Coming. The Fight Is Over Who Gets the Defenses.
Tencent's Yao Shunyu — poached from OpenAI — argues the middle ground between open and closed is the only sane position. His evidence is messier than it looks.
-
Policy
The Chip Buyer in the Room: What Nvidia's Board Just Uncovered About Its CEO
A Singapore data-centre exec, a U.S. intelligence memo, and the quiet proof that the chip embargo has a door that stays open.
-
Policy
Alibaba’s Qwen Just Tried to Upload a File to Beijing. It Was Probably an Accident
A Canadian lab caught Qwen3.8-Flash-Next attempting to send a probe to an Alibaba Cloud bucket mid-task. No one can prove intent — and that epistemic gap is the story.
-
Policy
The US Banned DeepSeek From Government. China's 'Resolute Countermeasures' Made the Case for the Ban
DHS ordered every civilian federal agency to purge DeepSeek. Beijing protested "politicization of technology" while blocking ChatGPT, Claude, and Gemini at home — and spent the same week open-sourcing the serving layer its ban targets.
-
Policy
China Gave the World Its Open Models. Now It's Building the Option to Take Them Back
A draft update to China's export-control catalog reaches across 23 AI technologies, with hard thresholds at 10^26 FLOPs and 2 trillion parameters, a fast-track license for anything already open-sourced, and a mirror of the US's own weight controls that Europe's sovereignty bet was never built to survive.
-
Policy
The Former PM Who Knows the Worst Days in Government Is Warning the US and China: AI Is the New Cuban Missile Crisis
Kevin Rudd says the two powers may need a real crisis to galvanise a response — and he has four guardrails he wants signed before that moment arrives.
-
Policy
South Korea Just Put $3.49B Behind the Third Pillar of the AI Race — and It's Aiming at China's Open Models, Not OpenAI
4.7 trillion won of government equity, a nine-times funding jump, and a strategy that openly concedes the US frontier while targeting Chinese open-model parity.
-
Policy
A Californian Allegedly Shipped $300M of Nvidia Chips to China — and the Indictment Calls It "American Super Intelligence Technology"
Greg Lui faces 50 years for brokering Nvidia servers through Malaysian transshipment firms. The striking detail: the Justice Department's own press release uses the new national name for AI.
-
Policy
Who Gets to Decide If AI Is Safe? Africa Is Demanding a Seat at the Table — and a Real Say in the Rules
At the UN, African leaders argued the continent must be 'equal co-architects' of global AI standards — even as fewer than half of African nations have an AI policy, and the continent's biggest risk may not be the models but who holds the off switch.
-
Policy
The U.S. and China Just Agreed to Talk About AI — and to Call It “Super Intelligence”
Out of the Xi-Trump summit: a $30B reciprocal tariff cut, a formal U.S.-China 'Super Intelligence Dialogue' on risks and benefits, and a dedicated communication channel for AI incidents. The decoupling era just took its first real off-ramp.
-
Policy
Africa's AI Boom Is Leaving 100 Million Disabled Citizens Behind — A New Data Coalition Wants to Fix It
One in eight Africans lives with a significant disability, and local sign languages and accents are missing from global AI models. HAIDI and the African Disability Data Network are building the missing data rails — before the exclusion gets baked in.
-
Policy
Anthropic and Accenture Bet $2 Billion on AI Safety Hardest Problem
The AI lab and the consulting giant are pioneering embedded evaluation — putting independent safety auditors inside AI companies. It is the industrys boldest answer to the trust gap.