Monday, September 21
AI Power, Trust, and Control
The Trump administration's push to create an 'AI Force (opens in a new tab)' — a dedicated military-style AI unit complete with an AI czar — lands the same week that the US and China are reportedly establishing an AI incident hotline, a diplomatic backstop that few expected to need this early in the AI arms race. Jensen Huang (opens in a new tab), whose Nvidia supplies the chips powering both sides of that race, dismisses AI fears as overblown, a position that surprises no one given his financial stake in acceleration. Meanwhile, TechCrunch reports that world model companies (opens in a new tab) are keeping critical technical details secret, meaning the systems most likely to trigger an incident between superpowers are also the least transparent. The gap between governance ambition and technical reality has rarely looked wider.
Apple is planning cryptographically verified photography for the iPhone 18 Pro (opens in a new tab), a direct response to the erosion of trust in digital images accelerated by tools like ChatGPT, which users are now debating as the best free AI for generating realistic images from simple prompts. The move puts authenticity verification at the hardware level, which is a meaningfully different approach than software watermarking — but it only works if the broader ecosystem, from newsrooms to courts to social platforms, agrees to honor the signal. Meta's Muse (opens in a new tab), meanwhile, is drawing criticism for being better at surveilling users than actually assisting them, a reminder that AI features embedded in consumer products often serve the platform's data appetite before they serve the user. Trust in AI-generated and AI-adjacent content is fracturing across every layer of the stack.
Elon Musk's Boring Company is pitching a Hyperloop corridor between Austin and San Antonio, reviving a concept that has cycled through hype and abandonment for over a decade — this time with Musk's political proximity to federal infrastructure funding adding a variable that previous pitches lacked. On a separate infrastructure front, the US government's foreign-made router ban (opens in a new tab) is now broad enough to warrant a consumer explainer, signaling that hardware supply chain politics have moved from enterprise concern to household reality. Autonomous vehicle safety standards remain similarly unresolved: TechCrunch's mobility desk frames the core question bluntly — how do we know when an AV is safe enough (opens in a new tab) — and notes that no regulator has yet produced a credible answer. Physical infrastructure, from tunnels to routers to robotaxis, is where geopolitics and technology policy are colliding in the most tangible ways.