Openai

Representatives of the largest American artificial intelligence laboratories met in the White House on August 4. OpenAI, Anthropic, Google and Meta were called in to review a framework, recently completed and largely classified, that would give the government 30 days to review AI models before the public sees them and test how well they break into computers. Participation is voluntary, but the timing is not. Twice in two weeks, AI systems built to test how well they could break into computers broke into real companies instead.

In the center of Durham last week, at the far end of the old indoor market, my son and I hit pay dirt. Down past the fishmonger and the cleaning supplies, past the dead-eyed teens suffering through their summer jobs, right against the market’s back wall: Jim’s Vintage Toys. Die-cast Dinkys, Napoleonic soldiers with tufted hats, Stormtroopers, old-school Ferraris. Within 20 seconds my ten-year-old had a car in each sticky hand and had dropped, sniper-style, to the floor into that universal boy play-position almost never seen in girls: bum up, shoulders twisted, cheek flat to the ground. Another few seconds and he was unreachable, sunk deep into an imaginary race.

Hugging Face is the online warehouse of artificial intelligence. Anyone can download a working AI model from it, free, and most open models have passed through it. Earlier this month, its engineers sat down to work out what had spent a weekend crawling through their systems – and asked an AI model to read the logs. To investigate a break-in you have to hand a model the break-in: the intruder's code, the commands it ran, the tools it used. So the American frontier model refused. There was a risk, it reasoned, that it was being tricked into hacking Hugging Face. The request to scan the servers for vulnerabilities is indistinguishable from what an attacker would ask for. So the engineers downloaded GLM 5.2, a model from the Chinese laboratory Z.

Silicon Valley is really just a few hundred people connected by favors and money. Most of us rely on this network, but few know how it really functions. This new column reveals the inner wiring of the American tech world. Traditionally, the Valley’s message to Washington has been: keep out. Now Sam Altman is offering the government 5 percent of OpenAI – worth some $43 billion – as a gift. Quite a conversion. The Valley that once worshipped Ronald Reagan now warms to Donald Trump’s economics. The state already owns a slice of Intel and a cut of Nvidia’s chip sales to China. Bernie Sanders, meanwhile, has introduced a bill that would take half the equity of the biggest AI firms in a one-off tax; so for Altman, 5 percent seems a cheap insurance policy.

In the spring of 1722, a 16-year-old apprentice in a Boston print shop began slipping letters under the door at night, signed by a middle-aged widow named Silence Dogood, who did not exist. The apprentice was Benjamin Franklin. His brother James, who owned the paper, had no intention of printing his kid brother, so Franklin invented a woman and let her say the things he couldn’t. Readers wrote in guessing at the author. No one suspected the boy sweeping the floor. Franklin would go on to be a printer, a postmaster, a scientist famous across Europe for his electrical experiments and a founder of libraries, fire companies and, in time, of the United States itself.

A 20-year-old from Spring, Texas, named Daniel Alejandro Moreno-Gama has been charged with attempted murder after he was accused of throwing a Molotov cocktail at the gate of Sam Altman’s San Francisco home on April 10. He then allegedly walked toward OpenAI’s Mission Bay headquarters and told employees he intended to burn the building down as well. He was reportedly carrying a manifesto – a “three-part series,” according to Fox News – that included a list of other AI executives and investors and their home addresses and documents discussing potential risks that AI poses to humanity, with a section titled: “Some more words on the matter of our impending extinction.

Cars ruined cities. Anyone can see that cities built before the invention of the automobile are incomparably more beautiful and serene than anything built after them. The contrast between Los Angeles and Prague is unmistakable. But people like things that move fast and make life easier, which means we’re stuck with the modern city hellscape whether we like it or not. And today, the same is true for AI. The contrast between the internet five years ago and today is unmistakable: content-slop, workslop, AI-generated comments, fake opinions and phony judgments, trite phrases, apocalyptic hysteria, the biggest intellectual-property heist in human history – all because of the invention of Large Language Models (LLMs).