OpenAI Funds Research to Stop AI From Being Used to Make Bioweapons

OpenAI, the company behind the popular AI tool ChatGPT, announced Monday that it is funding research to help prevent artificial intelligence from being used to create biological weapons. The company selected 14 organizations from more than 400 applications to receive special grants. One of those groups is working to make it easier for safety experts around the world to share information about AI-related biological risks.

One of the groups chosen is called the Nuclear Threat Initiative, or NTI. It is a non-profit organization that works to prevent disasters caused by nuclear and biological weapons. NTI will use its grant money to build a system that helps international experts share information about the ways AI could be misused to cause biological harm. The goal is to make that information easy to find and use across different countries and labs.

Hayley Severance, a deputy vice president at NTI, explained why sharing information is so important. She said the project looks at how to reduce the risks of AI being used in biology, and how to make sure experts from different countries can talk to each other about those risks. Right now, different countries have different laws about using AI to change or design living things, and those differences make it hard for labs to share what they know. Severance said the project will look at the barriers that get in the way of sharing and find ways to break them down.

Scientists are already worried because researchers have shown that AI can design a new virus that does not exist in nature. That means people with bad intentions could use AI to create dangerous germs or other biological threats. Severance said that even skilled scientists who already have access to advanced tools can do even more harm as AI gets better at helping people understand and change biology. This makes monitoring how AI is used in biology more important than ever.

OpenAI said it has committed one million dollars total to the new round of grants. Each group that receives a grant can also get up to one million dollars in API credits, which are like tokens that let them use OpenAI's technology. NTI plans to use those credits to help label and organize information so that experts can study it more deeply and on a larger scale. OpenAI said it hopes the grants will help researchers turn big ideas into real, practical work that can actually make the world safer.