The Trump administration may think regulation is crippling the AI industry, but one of the industry’s biggest players doesn’t agree.
At WIRED’s Big Interview event on Thursday, Anthropic president and cofounder Daniela Amodei told editor at large Steven Levy that even though Trump’s AI and crypto czar David Sacks may have tweeted that her company is “running a sophisticated regulatory capture strategy based on fear-mongering,” she’s convinced her company’s commitment to calling out the potential dangers of AI is making the industry stronger.
The Big InterviewWIRED’s iconic series returned to San Francisco with a series of unforgettable, in-depth live conversations. Check out more highlights here.
“We were very vocal from day one that we felt there was this incredible potential [for AI],” Amodei said. “We really want to be able to have the entire world realize the potential, the positive benefits, and the upside that can come from AI and in order to do that, we have to get the tough things right. We have to make the risks manageable. And that’s why we talk about it so much.”
Over 300,000 startups, developers, and companies use some version of Anthropolic’s Claude model and Amodei said that, through the company’s dealings with those brands, she’s learned that, while customers want their AI to be able to do great things, they also want it to be reliable and safe.
“No one says ‘we want a less safe product,’” Amodei said, likening Anthropolic’s reporting of its model’s limits and jailbreaks to that of a car company releasing crash-test studies to show how it’s addressed safety concerns. It might seem shocking to see a crash test dummy flying through a car window in a video, but learning that an automaker updated their vehicle’s safety features as a result of that test could sell a buyer on a car. Amodei said the same goes for companies using Anthropic’s AI products, making for a market that is somewhat self-regulating.
“We’re setting what you can almost think of as minimum safety standards just by what we’re putting into the economy,” she said. “[Companies] are now building many workflows and day-to-day tooling tasks around AI, and they’re like, ‘Well, we know that this product doesn’t hallucinate as much, it doesn’t produce harmful content, and it doesn’t do all of these bad things.’ Why would you go with a competitor that is going to score lower on that?”
Amodei said Anthropic has become noted for its commitment to what it calls “Constitutional AI,” where it trains its models on a baseline set of ethical principles and documents that teach human values. Using something like the United Nations Universal Declaration of Human Rights to train a model, Amodei said, can quickly teach a LLM how to respond to queries based not on the idea that a query is right or wrong, good or bad empirically, but rather than an issue is right or wrong in an overall ethical sense.
Anthropic’s commitment to creating a better, more ethical AI model has also helped it retain talent, Amodei said. “The story that we hear from people that come in the door [at Anthropic] is there’s something about the mission and the values and this desire to be honest about both the good and the bad, and the desire to help to make the bad things better, that feels very genuine, like we mean it,” she explained.
Perhaps that’s why Anthropic has grown its staff by leaps and bounds over the past few years, from 200 staffers to over 2,000. While those numbers could seem scary, especially when considering all the AI bubble talk flying around Wall Street and Silicon Valley, Amodei said she hasn’t seen any sign of her company or industry slowing down.
“Based on what we’re seeing, the models are continuing to get smarter at the exact sort of curve that the scaling laws talk about and the revenue is continuing on that same curve,” Amodei said. “As any of the scientists that work at Anthropic would tell you, everything continues going on the curve until it doesn’t and so we really try to be self aware and humble about that.”
The post Anthropic’s Daniela Amodei Believes the Market Will Reward Safe AI appeared first on Wired.
