How do you get an AI to answer a question it’s not supposed to? There are many such “jailbreak” techniques, and Anthropic researchers just found a new one, in which a large language model can be convinced to tell you how to build a bomb if you prime it with a few dozen less-harmful questions … Read more
Photo: Tom Williams/CQ-Roll Call, Inc via Getty Images (Getty Images) Days after Sam Bankman-Fried received a 25-year prison sentence, the FTX founder released a statement to ABC News saying how sorry he was about his crimes. He says he made some “bad decisions” and didn’t realize that his billion-dollar Ponzi scheme wasn’t illegal. Will More … Read more
In this photo illustration, Anthropic logo is seen on a smartphone screen. Pavlo Gonchar | SOPA Images | Lightrocket | Getty Images Bankrupt crypto exchange FTX has struck a deal with a consortium of buyers to sell the majority of its stake in artificial intelligence startup Anthropic for $884 million, according to a filing submitted … Read more
Deep-pocketed, sovereign wealth funds are among the investors clamoring to get a stake in Anthropic, the red-hot artificial intelligence startup that’s taking on OpenAI. One country that’s being left out: Saudi Arabia. As bankers line up a group of potential new Anthropic backers, the company has ruled out taking money from the Saudis, according to … Read more
Anthropic on Monday debuted Claude 3, a suite of artificial intelligence models that it says are its fastest and most powerful yet. The new tools are called Claude 3 Opus, Sonnet and Haiku. The company said the most capable of the new models, Claude 3 Opus, outperformed OpenAI’s GPT-4 and Google’s Gemini Ultra on industry … Read more
While AI tools offer new capabilities for web users and companies, they also have the potential to make certain forms of cybercrime and malicious activity much more accessible and powerful. Case in point: Last week, new research was published that shows large language models can actually be converted into malicious backdoors, the likes of which … Read more
Most humans learn the skill of deceiving other humans. So can AI models learn the same? Yes, the answer seems — and terrifyingly, they’re exceptionally good at it. A recent study co-authored by researchers at Anthropic, the well-funded AI startup, investigated whether models can be trained to deceive, like injecting exploits into otherwise secure computer … Read more
Elon Musk’s artificial intelligence company xAI organized in Nevada as a for-profit benefit corporation, The Information reported on Tuesday. The company behind the Grok chatbot is the latest AI startup to give itself an unusual corporate structure that, on paper, prioritizes the public good alongside its profits, following in the footsteps of OpenAI and Anthropic. … Read more
Anthropic is reportedly in talks to raise $750 million from Menlo Ventures, which values the startup at $15 billion, according to The Information. Like OpenAI, Anthropic also has a complicated board structure with misaligned values, and this recent influx of cash begs the question: is Anthropic bound to be the next AI board showdown? Why … Read more
Elon Musk tweeted a link to a strange letter Tuesday, supposedly written by former OpenAI employees detailing “a disturbing pattern of deceit and manipulation by Sam Altman and Greg Brockman.” Altman and Brockman, OpenAI’s CEO and president, were fired and then rehired in a dramatic failed coup that played out over the last five days. … Read more