The autonomous vehicle industry's credibility problem got a double hit: Tesla admitted its robotaxis are sometimes piloted remotely by humans, while Baidu's fleet in Wuhan trapped passengers on highways during a system outage. The Milk Road AI take on all this is contrarian: Uber is the quiet winner here, having built partnerships with Waymo, Rivian, NVIDIA, and 15+ others so it profits regardless of which AV tech wins. Tesla's vertically integrated cost structure is real—rides averaging $8.17 in SF versus Lyft's $15.47—but that transparency gap on the "autonomous" claims is a problem Matt Levine would recognize: he argues whether a hyped tech pitch is securities fraud depends entirely on what was claimed versus caveated—using Exxon's failed $500M algae project as his case study.

OpenAI's $122 billion funding round is being read by Om Malik as an elaborate IPO setup rather than a genuine capital raise—much of the money is circular (Amazon's tied to AWS contracts, Nvidia's in compute not cash), and the credit facility from major banks is really an underwriting audition. The tell, Malik argues, is the use of ARK Invest ETFs to pre-seed retail demand before the company even files—and the quiet shutdown of Sora signals they're cutting costs to make the numbers look cleaner ahead of a public offering. Funny enough, that same ARK Invest shows up in both Malik's analysis and Milk Road's, where ARK's projections on robotaxi pricing are used to frame Tesla's cost advantage—suggesting ARK is actively positioning itself as a narrative infrastructure player across multiple emerging sectors.

Trump's week in courts was rough: a federal judge struck down his executive order defunding NPR as First Amendment retaliation, and separately blocked his $400 million White House ballroom project pending congressional approval—which ARTnews flags raises its own concerns, since the private funding comes from corporate donors who've received substantial government contracts. Reason is also flagging that Trump's 2016 promise to eliminate the national debt has aged badly—it's doubled in a decade—and the Supreme Court is now weighing birthright citizenship challenges in Trump v. Barbara. Meanwhile, Iran has threatened attacks on Apple, Google, and Microsoft, and WIRED reports the US military's GPS software is a decade behind schedule and $8 billion over budget—a fun pairing with Reason's piece on US military buildup in Israel.

Conservative advocacy groups are using AI book-scanning tools like BLOCKADE and NarraTrue to automate school library challenges—the AI outputs are formatted to resemble legitimate critiques and submitted as formal evidence, disproportionately targeting LGBTQ+ titles, while a company called BookmarkED is selling compliance products to the very districts being pressured. The tools are known to be unreliable—one advocate documented that AI tends to over-apply legal standards—but Utah's bright-line rule has already gotten 1,400+ books pulled in one district. Which brings up the Reason item about a Tennessee grandmother who spent six months in jail after faulty AI analysis—a different context, same throughline: AI tools are moving faster than the accountability structures around them.