08/08/2026
🚨 Gates-funded AI creates 16 never-before-seen viruses that can copy themselves in the lab
Scientists at Stanford and the Arc Institute used special AI programs called Evo 1 and Evo 2 to write complete sets of genetic instructions for viruses.
Those digital instructions were then turned into real DNA, assembled, and placed inside common lab bacteria known as E. coli, where sixteen of them successfully created living viruses that could infect the bacteria, make more copies of themselves, and keep spreading.
This is the first time AI has written entire working virus genomes that scientists then built and proved in the lab.
They began with a well-known natural virus called ΦX174 that only attacks bacteria.
The AI generated hundreds of thousands of possible new versions, and researchers manufactured and tested nearly 300 of them.
The sixteen that worked were clearly different from anything found in nature, with each one containing dozens to hundreds of changes in its genetic code and thirteen including brand-new mutations never seen before.
One even used a key building-block protein from a completely different virus that earlier scientists had not been able to get to work.
Several of the new viruses spread faster and more strongly than the original, and when used together they quickly overcame bacteria that had become resistant to normal versions.
These viruses only attack bacteria and cannot infect people, and the AI was never trained on any human viruses.
Even so, this remains high-risk research with serious unknown dangers.
AI can now write a full virus genome, companies can turn that writing into actual DNA, and the result is a virus that can copy itself.
We also do not know anything about the longer-term consequences of this kind of genetic engineering once these novel sequences exist in the real world or how they might change over time.
The main researcher, Brian L. Hie, who planned and led the work, has reported receiving support from the Gates Foundation, an organization with a long record of funding high-risk biological research that many view as reckless and dangerous.
The scientists and outside security experts say this raises urgent safety questions because the systems that check DNA orders for dangerous material mainly look for known natural threats and may miss these new AI-made ones.
Any possible medical uses against hard-to-treat bacterial infections do not reduce the real risks of accidental release, deliberate misuse, or unintended effects that could prove difficult or impossible to control.
The involvement of Gates-linked funding only heightens those concerns.
Clearer rules and stronger protections are needed before this technology spreads more widely and falls into the wrong hands.
