The challenge
AI is expanding by leaps and bounds, infused into everything from chat platforms to how companies hire employees.
But as AI grows, so do the risks. Models can be wrong. They can take in misinformation and perpetuate stereotypes and bias. They can hallucinate and provide incorrect or even dangerous answers. AI is also being used by scammers to do everything from making fake versions of real people to stealing sensitive information, or creating illegal and offensive images.
THE SOLUTION
In 2021, Hertz Fellows Dario Amodei and Jared Kaplan co-founded Anthropic, an AI safety and research lab. The company collaborates with civil society, government, academia, nonprofits and industry to promote AI safety across the field.
They aim to build safer AI systems — ones that are reliable, interpretable and steerable. To achieve that goal, they take a scientific approach to AI, conducting research to develop and apply safety techniques, and then apply those techniques in partnerships with AI companies.
“Our existence in the ecosystem hopefully causes other organizations to become more like us. That’s our general aim in the world and part of our theory of change.”
Dario Amodei
Co-Founder and CEO, Anthropic
THE IMPACT
Anthropic has become a leader in an AI-safety movement. In 2023, they collaborated with the AI Safety Institute, and in 2024, more than 20 companies joined Anthropic in support of stopping the deceptive use of AI in 2024 elections.
Anthropic was also the first company to use Constitutional AI, where its LLMs are given a set of principles to discourage and mitigate broadly harmful outputs. It was also the first to establish a form of voluntary self-regulation called Responsible Scaling Policy, which is a series of technical and organizational protocols to help the company manage risks faced when developing increasingly capable AI systems.
“Our existence in the ecosystem hopefully causes other organizations to become more like us,” Amodei said. “That’s our general aim in the world, and part of our theory of change.”
