Summary
A study shows that artificial intelligence can design DNA for dangerous proteins in ways that biosecurity checks might miss. Researchers used AI to generate DNA codes for harmful proteins that evaded the detection systems of DNA manufacturers. The study raises concerns about the potential misuse of AI in biotechnology.
Key Facts
- The study was published in the journal Science.
- AI was able to design DNA codes for more than 75,000 hazardous protein variants.
- The biosecurity screening systems failed to detect some of these AI-generated sequences.
- DNA manufacturers have added fixes to their screening software, but some variants still go undetected.
- The study highlights ongoing worries about AI being used to create biological threats.
- Researchers and the journal limited the distribution of certain details from the study to prevent misuse.
- They enlisted a third-party nonprofit to control access to sensitive information.
- This approach is a first in managing the risk of sharing hazardous data in scientific publications.
