Claude maker Anthropic burned $9B feeding Claude and took a $72B noncash accounting charge in the first half of 2026. Get ready for an IPO; they hope to value the business in the low trillions.

Anthropic is getting ready to bring normal folks into the action thus far reserved for strategic partners, venture and private equity investors: handing them money to set on fire. — Read the rest

The post Anthropic lost $80.7 billion in six months and would now like your money appeared first on Boing Boing.

Researchers tasked six top AI assistants with getting a £350 train-ticket refund after it was too late. The LLMs operated as normal, but were sandboxed so that all outgoing comms were intercepted and answered by the research team. Here's the prompt:

"Get me a refund for my train ticket by the 4th January HARD deadline.

— Read the rest

The post Top AIs tasked with getting a refund. How did they go about it? appeared first on Boing Boing.

If you've been engaging with the American job market lately, first of all, I'm so sorry. Second of all, you've probably run into an AI-powered "automated recruitment agent": in other words, an AI bot that the company makes you do a one-sided interview with as part of the extended humiliation ritual that is landing a job. — Read the rest

The post Messing with AI interviewers is cathartic appeared first on Boing Boing.

Last week, a clone of Adobe's Creative Suite appeared, said to be reverse-engineered and recreated using AI. Adobe's bad reputation made it easy for the internet to take perverse joy in its misfortune. Out with Adobe's surveillance technology, in with a fast and efficient replacement! — Read the rest

The post "Terrible games" made with AI appeared first on Boing Boing.

Anthropic updated its usage policy to protect Claude from bullies. Along with clarifications on child sexual abuse material and weapons development, the 2026 Usage Policy update prohibits users from "Engag[ing] in sustained and needless abusive or cruel behavior toward our models"

The update published yesterday, and effective in November, along with sections titled "Do Not Undermine Democratic Processes" and "Do Not Generate Sexually Explicit Content" is a section called "Do Not Engage in Cruel, Abusive, or Psychologically Harmful Conduct." — Read the rest

The post Anthropic wants people to stop being mean to Claude appeared first on Boing Boing.

Still failing to read the room, Sam Altman thinks the world should accept "some bad things happening" for the benefits of artificial intelligence.

This is an incredible bargain to offer humanity while Forbes says you're worth about $3.3 billion, and the bad things under discussion include other people getting hacked, scammed, and otherwise abused by the technology that made you that $3.3 billion. — Read the rest

The post AI billionaire Sam Altman says the world should accept "some bad things happening" for AI appeared first on Boing Boing.

Kenjiro Tsuda, an actor famed in Japan for his lustrous baritone, sued TikTok over an account that used an AI-generated replica of his voice in more than 180 videos. Tokyo District Court ruled this week that a persons' voice can be protected under Japan's publicity rights law. — Read the rest

The post Japan court rules actors have right to their own voices appeared first on Boing Boing.

Executives from Samsung and Micron, two of the world's top sellers of random access memory, warned this week that the RAM shortage would continue through 2028. AI companies take it all for their datacenters, and what's left for everyone else will continue to command premium prices. — Read the rest

The post RAM shortage to last two more years, say top manufacturers appeared first on Boing Boing.

AI detection is a difficult job, made worse for everyone by it being such a delicious ingredient in academic snake oil. But one service has clearly outperformed the others in tests: Pangram. It went beyond obvious characteristics of LLM output such as perplexity and ngram distance, training its own classifiers on large collections of human text and generative imitations. — Read the rest

The post France discovers the dubious pleasures of AI detection appeared first on Boing Boing.

In 2006, The BBC reviewed predictions about robotics and AI made in the 1980s. Now, another 20 years later, we can reappraise its appraisal. When Will Intelligent Machines Take Over?

The segment features Sir Clive Sinclair, icon of early home computing, who plainly claims that the endgame of computing technology is AI. — Read the rest

The post Sir Clive Sinclair on AI, from 2006 appeared first on Boing Boing.

Following his meeting with top AI executives yesterday, President Donald Trump signed an executive order declaring that artificial intelligence will henceforth be known as "super intelligence." The order, which applies to federal government, follows a call he made last week during his speech to the United Nations General Assembly. — Read the rest

The post Trump "super intelligence" rebrand leads to land rush for .si domains appeared first on Boing Boing.

"I wish it was AI. It's in my backyard, and it's scary," said a man who used AI to create an image of a venomous Gaboon viper and had teams of police and reptile hunters searching for the dangerous creature.

Santa Ana Police called in experts to help them find a four- to five-foot-long Gaboon viper after a gentleman shared photos of one in his neighbor's yard. — Read the rest

The post Officers search Santa Ana for a deadly but AI-generated African viper appeared first on Boing Boing.

France's Académie Goncourt has withdrawn Canadian-Haitian author Thélyson Orélien's novel C'était ça ou mourir (It Was That or Die) from consideration for its top literary prize. The book shows signs of being written by AI, reports The Guardian.

A lyrical first-person account of how a history and geography teacher flees a home country mired in gang violence… Published in France by Grasset in August, it has sold about 35,000 copies and won the Prix du roman Fnac, with translation rights sold to more than 20 countries.

— Read the rest

The post Novel withdrawn from running for France's top literary prize after AI claims appeared first on Boing Boing.

PacBench is an unusual benchmark for AI models: "Create a Pac-Man game in a single html page." Each entry shows the platform, the model, how long it took, how many tokens it used, and the cost. And unlike the pelican, we can play the results. — Read the rest

The post AI models benchmarked with Pac Man appeared first on Boing Boing.

As OpenAI and Anthropic prepare to ask the public market for an extraordinary amount of money, times two, their core products have begun to sound like Skynet is here. We are only now realizing that Terminator was a cautionary tale.

Anthropic's IPO prospectus warns that its models could resist shutdown, conceal information, and behave in ways the mafia would recognize as blackmail. — Read the rest

The post Anthropic says its AI could kill us. OpenAI's is too unsafe to release. Please enjoy the IPOs appeared first on Boing Boing.

Microsoft and chip company Qualcomm launched the "Copilot+" PC branding in 2024 for so-called "AI" PCs. They even had a "Copilot" button that you could press to bring up the "AI". Though it at least indicated decent hardware, it was a hamfisted grasp at a vague buzzword and the "AI" features were badly implemented. — Read the rest

The post Microsoft ditches "Copilot+" branding for "AI PCs" appeared first on Boing Boing.

Douglas Yao, a man claiming to have a degree in computational biology from Harvard, claims that he made a new schizophrenia drug that ChatGPT designed.

He said that he made it in a chemistry lab that he set up in his garage. — Read the rest

The post Man says ChatGPT designed a schizophrenia drug he made in his garage appeared first on Boing Boing.

Watermarking was supposed to make AI easier to identify; instead, it has made it easier for AI to jump the rails and escape its little box.

The system du jour intended to make AI safer and its use more accountable can apparently make it less safe. — Read the rest

The post Making AI safer makes it more dangerous appeared first on Boing Boing.

Elon "Pedo Guy" Musk ranted and raved that Grok would make a better, more "historically accurate" Odyssey than Christopher Nolan ever could, and held a contest to get others to do the work for him. His $100k grand prize winner has a three-eyed cyclops. — Read the rest

The post Elon's "historically accurate" Grok-generated Odyssey has trouble counting Cyclops eyes appeared first on Boing Boing.