Antrophic, an AI company founded as recently as 2021, has received more than half a billion US dollars from investors - without a real business model. What makes the startup so special?
Many companies are researching and developing AI systems that will be or are already being used on a large scale in the future. Often, this involves fundamental tasks that can have a major impact on our society: how we get around, exchange information, make consumption decisions, and what professions we pursue. AI potentially permeates all areas of our lives.
Accordingly, the reliability, security, and accountability of large-scale AI systems are of great importance. If AI is to realize its full potential, it must persist in these fields. Conversely, failures could shake confidence in the technology, thus slowing down development or - worse - having serious consequences for humanity.
$704 million for AI security
AI Startup Antrophic specializes in AI security research. It says it is working on a "large-scale experimental infrastructure" to research and improve the security of large-scale AI models with the goal of making AI systems "more steerable, robust and interpretable".
Since its founding in 2021, Antrophic has already received $704 million from Investors. In the second round of funding now announced, 580 million US dollars will be added.
The fact that such large sums flow into a research startup can probably be explained by the relevance of the topics mentioned and a partly altruistic motivation of the investors.
The investment round is led by Sam Bankman-Fried, the founder and CEO of cryptocurrency exchange FTX. His net worth is estimated at more than $25 billion.
The Swiss-based Center for Emerging Risk Research (CERR) is also a new addition to Antrophic's investor list. It says its goal is to "improve the quality of life of future generations as comprehensively as possible" and to promote research and implementation of "solutions to avoid technological risks." Artificial Intelligence is a focus area for CERR.
Former Google CEO Eric Schmidt is already one of Antrophic's investors. Among other things, Schmidt established an altruistically motivated grant program in early 2022 for researchers contributing to the goal of powerful, safe general AI.
There's a lot of OpenAI in Anthropic
Antrophic was founded by siblings Dario and Daniela Amodei. Dario Amodei previously worked as a lead senior researcher at OpenAI for several years. For two years he was responsible for the work of OpenAI's AI security team.
Daniela Amodei was responsible for security and policy at OpenAI from May to December 2020. Prior to that, she also worked at OpenAI as an engineer. The siblings were also involved in the development of OpenAI's GPT-3. Together, they spent nearly six years on OpenAI's research teams.
Similar to OpenAI, Antrophic is motivated partly by science and altruism, pursuing commercial interests as well, but without a fixed business model. Currently, the startup is focused primarily on researching security targets, according to its self-description: "Down the road, we foresee many opportunities for our work to create value commercially and for public benefit."
The newfound capital will go toward expanding the company's machine learning infrastructure. The startup plans to train large AI models and then analyze them for security features to better understand what makes secure, integrated AI models work.
“We’ve made strong initial progress on understanding and steering the behavior of AI systems, and are gradually assembling the pieces needed to make usable, integrated AI systems that benefit society,” CEO Dario Amodei says.
Currently, Antrophic has about 40 employees and plans to grow further. The headquarters are located in San Francisco.