AIpocalypse.Now
Today'sdoom4.0
Doom 4/10 · 8 stories

UnitedHealth Sics AI on Doctors as Regulator Quits

An insurance giant automates the phone tree from hell, a UK watchdog tells the wrong joke, and a Sam Altman biopic dies in development.

Published · By · Story-level doom average 2.9/10

The Insurer Will See You Now, Via Bot

UnitedHealth is spending billions to put AI on both ends of the healthcare phone call. Bots will ring doctors. Bots will read charts. Bots will monitor the human agents who are still, for now, employed. Bots will schedule the appointments that other bots will later deny coverage for. This is the most important story in the slot and it is being reported with the placid tone of a quarterly earnings note.

The pitch is efficiency. The reality is that the single largest US health insurer is industrializing the part of the system that already grinds patients and physicians into paste. When a denial comes from a model trained on prior denials, the appeal process becomes a Turing test you did not sign up for. Doctors will spend their days arguing with software that has been optimized to outlast them. Some will give up. That is the feature, not the bug.

Nothing about the technology here is novel. What is novel is the willingness to deploy it at the scale of a country's medical infrastructure before anyone has agreed on what an AI-generated coverage decision legally is. Regulators are not ready. The doctors are not ready. The bots are extremely ready.

Watchdog Tells Joke, Exits Stage Left

The UK's Information Commissioner has resigned after inappropriate humor landed badly, ending a tenure that began in January 2022. The timing is unhelpful. Britain is mid-scramble on AI rules, the office sits at the intersection of data protection and model governance, and the chair is now vacant because someone could not read a room.

Across the Atlantic, Trump declared Anthropic poses no security threat, while his administration restricts Anthropic's model access in practice. This is the new bipartisan style of AI policy, where the press release and the procurement memo describe different countries. Anthropic gets to be both safe enough to praise and dangerous enough to throttle, depending on which official is talking and which contract is at stake. The company's lobbyists are presumably billing by the contradiction.

Taken together, the regulatory picture is a vacancy in London and a shrug in Washington. The companies notice.

Research Promises, Hollywood Passes

A startup called Subquadratic emerged from stealth claiming it has cracked the computational bottleneck that has defined large language models since the transformer. The claim is that you can get past the quadratic attention cost without losing the capabilities. Every six months someone announces this. Occasionally one of them is right. MIT Tech Review is doing its job by reporting the claim; the rest of us should wait for a reproduction before rearranging the data center.

If it works, the economics of inference change overnight, which means the economics of everything downstream change too, including the UnitedHealth bots. If it does not, it joins the long list of architectural Hail Marys that briefly inflated a Series A.

Meanwhile, Amazon has killed the Sam Altman biopic. Luca Guadagnino was directing, Andrew Garfield was attached, and the script was reportedly built around the November 2023 board ouster and the weekend that followed. After a year of development, Amazon walked. The official reasons are unofficial. The unofficial reason is that nobody wants to be on the wrong side of OpenAI's media relations team while also competing with them in cloud and models. The story of the board firing Altman remains untold by Hollywood, which is itself a kind of ending.

The Trillionaire Problem

Elon Musk's net worth is approaching a trillion dollars on paper, propped up by a pending SpaceX listing and the AI valuation surge lifting Tesla and xAI. Americans surveyed about this are not impressed, they are worried, because their retirement accounts are now structurally exposed to one man's mood. Index funds were supposed to be the safe choice. They are now a leveraged bet on whether Musk sleeps.

Elsewhere in the cycle, a Guardian cartoonist is mining 2026 for fresh insults, and a group of seniors in Tulsa have held their Wii bowling title for six years running. The Okies are undefeated, the controllers still work, and somewhere in a nursing home rec room the machines are being used exactly as intended. Hold onto that image. You will need it tomorrow.

Sources cited in this digest
  1. UK AI Regulator Exits After Jokes Land Wrong · BBC News · 2/10
  2. Trump Declares Anthropic Not Threat; Actions Suggest Otherwise · · 4/10
  3. Senior Citizens Dominate Virtual Bowling With Six Year Streak · New York Times · 1/10
  4. Columnist Finds 2026 Insult Material in Daily News Cycle · The Guardian · 1/10
  5. UnitedHealth Deploys AI Bots to Call Doctors, Handle Claims · · 6/10
  6. Amazon Drops Sam Altman Biopic Starring Andrew Garfield · The Verge · 2/10
  7. Startup Claims Breakthrough in Large Language Model Bottleneck · MIT Tech Review · 4/10
  8. Musk's Trillion Dollar Fortune Worries Americans About Retirement Funds · The Guardian · 3/10
Today's doomWeekly column