Closing Time in Musk v. Altman
Closing arguments in Musk v. Altman landed today, and if you were hoping for a dignified legal reckoning over the future of AI, I have bad news. Musk’s attorney Steven Molo apparently called co-defendant Greg Brockman “Greg Altman,” incorrectly stated Musk wasn’t seeking damages, and generally delivered what reporters described as a demolition derby — of his own case. The jury is now deliberating on what Wired correctly notes has made pretty much everyone involved look bad. When the most coherent narrative coming out of a billion-dollar AI lawsuit is “both sides lost the PR war before the verdict,” you have to wonder what exactly we’re doing here.
Behold, the Jackass Trophy
Before the jury was seated, Sam Altman’s team passed what appeared to be a little league trophy to the judge — who then made the lawyers read the inscription aloud for the press: “Never stop being a jackass.” It was a gag gift OpenAI employees bought for Musk years ago and apparently decided this was the moment to deploy it. I genuinely cannot tell if this is genius courtroom theater or the most expensive trolling in tech history, but either way, someone at OpenAI has excellent comedic timing.
Microsoft Cancels Its Own Claude Code Experiment
This one is quietly significant: Microsoft, which gave thousands of its own employees access to Anthropic’s Claude Code starting in December, has begun canceling those licenses. The program was designed to get non-engineers — project managers, designers — experimenting with code for the first time. Sources say it actually worked, which makes the cancellation all the more curious. Microsoft owns a substantial chunk of OpenAI, ships Copilot in everything it makes, and apparently decided having its own employees enthusiastically using a competitor’s coding tool was a bit on the nose.
OpenAI Codex Comes to Your Phone
Hot on the heels of Microsoft’s quiet retreat from Claude Code, OpenAI is pushing Codex into the ChatGPT mobile app on iOS and Android. The timing is not subtle — Claude Code has been the coding tool everyone’s been talking about, and OpenAI is moving fast to close the gap, including reportedly cutting internal “side quests” to focus resources. Codex on your phone is a genuinely useful idea; the ability to monitor and steer long-running coding tasks from anywhere could change how developers actually work. Whether it’s enough to shift momentum back from Anthropic is the real question.
Anthropic Launches Cowork — Claude Code for the Rest of Us
While OpenAI is racing to catch Claude Code, Anthropic is already moving up the stack — Cowork is a new Claude Desktop agent that brings agentic file-working capabilities to non-technical users, no coding required. The kicker: the team apparently built the whole feature in about a week and a half, largely using Claude Code itself. That’s either a remarkable proof-of-concept for agentic development or the most recursive product launch in Silicon Valley history. Probably both.
AI Medical Scribes Are Getting Drugs Wrong. A Lot.
Ontario auditors evaluated AI scribe systems being used to transcribe doctor-patient conversations into medical notes and found that 60% of them mixed up prescribed drugs. Not occasionally — routinely. This is the kind of story that should be front-page news but tends to get buried under trial drama and product launches. AI in healthcare has enormous potential, but “it confidently wrote down the wrong medication” is a category of failure that has real consequences for real patients. The gap between demo and deployment has never been more important to take seriously.
OpenAI Is Apparently Preparing to Sue Apple
OpenAI is reportedly exploring legal action against Apple over a ChatGPT integration that didn’t deliver the subscribers and prominence OpenAI expected. This is a fascinating development — OpenAI’s deal to be baked into Apple Intelligence was supposed to be a massive distribution win, and apparently the reality didn’t match the pitch. OpenAI has now found itself in legal disputes or near-disputes with Elon Musk, a former board member, and now its most prominent consumer hardware partner. At some point, the pattern becomes the story.
SpaceXAI Is Bleeding Talent
More than 50 employees have left the newly merged SpaceXAI since February, according to TechCrunch. The reasons include the usual suspects — burnout, leadership changes, talent poaching — but one detail stands out: liquidity events may have weakened retention incentives by letting early employees cash out. Building a competitive frontier AI lab is hard enough without your best people deciding they’ve made enough money to go work somewhere less chaotic. grok continues to ship, but the organizational turbulence at xAI has been a constant background hum for months now.
The AI Poop App Tried to Sell Its Users’ Stool Database
I debated including this. I’m including it. An AI app that analyzes stool photos for health insights apparently offered 404 Media access to a database of 150,000 user-submitted poop images. The founder’s pitch: “I hoarded a large database of something valuable, just not what you expect.” This is either a privacy nightmare, a cautionary tale about health data monetization, or proof that the AI gold rush has reached a frontier that no one asked for. Possibly all three.
Bottom Line
From a courtroom jackass trophy to AI scribes prescribing the wrong drugs, this week’s AI news is a useful reminder that the technology’s biggest challenges aren’t compute or benchmarks — they’re accountability and trust.