Gates Warns That AI Could Enable Bioterrorism as His Own Funding Helped Build the Technology That Makes It Possible
Bill Gates is suddenly sounding the alarm about artificial intelligence, and this time his warning is considerably darker than the usual predictions about chatbots taking your job or teenagers using AI to write college essays.
In a recent CNN interview with Anderson Cooper, the Microsoft co-founder warned that artificial intelligence is becoming capable of creating serious bioterrorism risks, along with cyberattacks and other threats. Gates has argued that AI is advancing so quickly that governments may not have enough time to establish meaningful safeguards before the technology becomes extraordinarily difficult to control.
That warning would be considerably easier to accept at face value if Gates’ enormous philanthropic footprint wasn’t also tied to the very field that is making some of these frightening possibilities increasingly real.
Because while Gates is warning the world about AI-assisted bioterrorism, researchers have just demonstrated something that would have sounded like science fiction a few years ago: artificial intelligence can design complete viral genomes that scientists can physically synthesize into functioning viruses.
The research, published in the prestigious journal Science on August 6, 2026, involved scientists at Stanford University and the Arc Institute using genome-language models to design entirely new bacteriophages, viruses that infect bacteria. The researchers generated hundreds of candidate genomes, synthesized roughly 300 of them, and discovered that 16 produced viable viruses capable of infecting and killing E. coli.
Let that sink in.
This wasn’t simply an AI drawing a picture of a virus. It wasn’t a computer simulation. The AI generated genetic instructions. Scientists then turned those instructions into physical DNA, assembled the genomes and tested them in the laboratory. Sixteen of the AI-designed genomes actually worked.
The researchers were working with bacteriophages rather than viruses capable of infecting humans, and the study was aimed at developing new weapons against antibiotic-resistant bacteria. That is a legitimate scientific objective.
But it doesn’t erase the significance of what was demonstrated. AI can now help design living, self-replicating biological entities.
And that is precisely where Gates’ warnings about AI-powered bioterrorism become difficult to separate from the technological ecosystem that he and his foundation have spent years helping finance.
Gates Is Warning About the Fire While Investing in the Matches
The Gates Foundation has provided funding to Brian Hie, the Stanford and Arc Institute researcher whose laboratory is behind the work. In the earlier version of the research, Hie explicitly acknowledged Gates Foundation funding.
There is an important distinction here. The final Science paper does not list the Gates Foundation among the direct funders of the published 16-phage experiment. Stanford’s announcement instead identifies funding from Arc Institute, the National Science Foundation, the Knight-Hennessy Graduate Scholarship Fund, the Fannie and John Hertz Foundation and Stanford’s Institute for Human-Centered AI.
But the Gates connection to the underlying technology is real.
The broader Evo research that enables AI-driven genomic design has received Gates Foundation support. The Nature paper describing Evo 2 specifically acknowledges Gates Foundation funding for Hie. Evo 2 is a massive genomic language model trained on approximately 9 trillion DNA base pairs and designed to predict and generate biological sequences.
So the more accurate criticism isn’t that Bill Gates personally wrote a check labeled “MAKE 16 VIRUSES.”
The criticism is considerably more uncomfortable:
Gates has financially supported research in the ecosystem developing increasingly powerful AI systems for biological design, while now warning that AI could eventually facilitate bioterrorism.
That’s a contradiction worth examining.
The Technology Has Crossed a Line
For decades, genetic engineering required scientists to work painstakingly with existing biological material. The possibilities were enormous, but the process required specialized expertise, laboratories and considerable trial and error.
Generative AI changes the equation.
The researchers behind the Science paper demonstrated that genome-language models can explore biological possibilities at a scale humans simply cannot match. The system generated large numbers of candidate genomes and helped researchers identify sequences worth physically building and testing.
Sixteen of approximately 300 synthesized candidates proved viable.
Some of the resulting phages also displayed characteristics that made them biologically interesting. Researchers found that several could overcome bacterial resistance to the original virus, while some performed better than the naturally occurring template in laboratory experiments.
That is impressive science. It is also precisely why it should make people uncomfortable.
The important breakthrough isn’t that someone created a new way to kill E. coli. Humans have been manipulating viruses and bacteria for a very long time.
The breakthrough is that AI can dramatically accelerate the search through biological possibilities. And once you establish that a machine can design a functioning viral genome, the obvious question becomes: what happens when somebody applies increasingly powerful models to increasingly dangerous organisms?
That is not some ridiculous conspiracy theory. The authors themselves acknowledge the broader implications, and researchers at Johns Hopkins published a Science commentary alongside the study warning that the generation of functional viral genomes has urgent biosafety and biosecurity implications.
Gates’ Warning Is Right. His Position Is Still Deeply Troubling
This is where the hypocrisy becomes difficult to ignore. Gates is now warning that AI could eventually allow bad actors to conduct biological attacks.
He isn’t wrong. In fact, he may be more right than many of the people dismissing the concern.
The problem is that Gates is not an outsider watching this technological revolution from the sidelines. He is one of the world’s most influential technology billionaires and philanthropists. Through the Gates Foundation and related investments, his organization has poured enormous sums into biotechnology, infectious disease research, genomic science and AI-enabled biological research.
That gives him an enormous influence over which technologies get developed. You cannot spend years helping build increasingly powerful technological systems and then act shocked when those systems become powerful enough to create risks that were obvious from the beginning.
It is like helping construct a nuclear reactor and then giving a television interview warning everyone that nuclear reactors can be dangerous. The warning can be completely correct. The irony doesn’t disappear.
The Real Problem Is the Race to Build First and Regulate Later
The most disturbing part of the story isn’t Bill Gates himself. It’s the larger philosophy governing Silicon Valley, biotechnology and artificial intelligence.
Build it. Then demonstrate it. Publish it. Then commercialize it. Then start asking whether perhaps somebody should have thought about the consequences.
That sequence has become depressingly familiar.
AI researchers have already demonstrated systems capable of designing proteins, manipulating biological sequences and generating increasingly sophisticated genomic information. Evo 2 represents another significant step by applying language-model technology to DNA across an enormous biological dataset.
Now researchers have demonstrated that AI-generated genomes can become functioning viruses. Today they’re bacteriophages. Tomorrow the technology will be more capable. That doesn’t mean tomorrow’s AI automatically produces a human pandemic. It doesn’t. But pretending that the capability itself isn’t a major security issue would be reckless.
And once the information is published, the knowledge doesn’t magically disappear.
Gates Wants Regulation, Perhaps He Should Start With His Own Industry
Gates has called for stronger oversight of AI and has warned that governments need to move quickly before the technology gets too far ahead of regulators. He has even suggested international structures resembling the systems created to monitor nuclear technology.
That is a reasonable position. But it also raises an uncomfortable question:
Why wasn’t the same urgency applied before AI and synthetic biology became capable of doing this?
The answer, unfortunately, is the same answer that seems to govern almost every technological revolution. The upside gets monetized immediately. The downside gets discussed later. And the people who helped create the technology suddenly become the people explaining why everyone else needs to be careful with it. There is something profoundly backwards about that.
The 16 Viruses Should Be a Warning, Not a Celebration
The scientists deserve credit for demonstrating what AI can accomplish. The problem is that what they accomplished is simultaneously a warning about what AI can accomplish. The Science study proves that artificial intelligence has moved beyond merely analyzing biology. It can now participate in creating novel biological systems that actually function in the physical world.
That is an extraordinary technological achievement. It is also an extraordinary responsibility. Bill Gates is right to warn that AI could eventually contribute to bioterrorism.
But if he genuinely believes that threat is as serious as he now claims, the public has every right to ask why the technological ecosystem surrounding AI-driven biological design has received so much investment and encouragement from institutions connected to him.
The answer cannot simply be, “Don’t worry, we have safeguards.”
History has a rather annoying habit of demonstrating that safeguards designed by the same people racing ahead of everyone else don’t always remain safeguards for very long. The genie is already out of the bottle.
AI can write code. It can design proteins. It can analyze genomes. And now it can help design complete viral genomes that scientists can turn into functioning biological organisms. Bill Gates is warning us that this could someday become a weapon.
The uncomfortable part is that the world doesn’t need to wait for someday to see how close the technology is getting. The experiment has already been published. And 16 of those AI-designed viruses actually worked.
