Article By Frank Bergman
A team of California scientists has just made a chilling announcement after intentionally infecting a human patient with an artificial intelligence-developed virus.
San Francisco-based AI research lab Radical Numerics says it has administered an artificial intelligence-engineered viral treatment to a human patient.
However, the lab admits that it infected the patient with the virus without publishing peer-reviewed findings or releasing clinical details for outside scrutiny.
Radical Numerics announced the experiment Tuesday in a technical report on its website, describing how it used AI to develop a treatment targeting a bacterial infection.
The announcement takes AI-designed biology beyond laboratory experiments and into human treatment, while leaving major questions about the patient’s outcome unanswered.
The viruses involved are bacteriophages, which infect bacteria.
The treatment is intended to attack harmful bacteria, rather than infect human cells.
But the company’s decision to withhold clinical information has drawn skepticism from outside scientists.
Human Treatment Announced Without Peer Review
Phage therapy uses bacteria-targeting viruses to combat infections.
Finding a virus capable of attacking a particular bacterial strain, however, can be difficult and time-consuming.
Radical Numerics says AI can transform that process.
Introducing Omnii-1, our most powerful genome language model, and a major step toward general biological intelligence In a world first, Omnii-1 designed the first AI-generated bacteriophages to be administered to a human patient with an antibiotic-resistant infection under compassionate use
— Eric Nguyen (@exnx) October 20, 2026
Its announcement was self-published and has not been peer-reviewed.
The New York Times describes the material as closer to a blog post than a conventional scientific study.
The company declined to provide the newspaper with clinical information about the patient or identify the doctors involved, citing the proprietary nature of a treatment it hopes to commercialize.
“We take our responsibility to protect patient confidentiality and maintain the trust of our clinical partners extremely seriously,” Radical Numerics CEO Eric Nguyen told the Times.
“We intend to publish our scientific findings through peer-reviewed research, with any future disclosure of clinical details subject to appropriate consent,” Nguyen added.
Harvard microbiologist Michael Baym told the newspaper that a successful treatment would be remarkable, but expressed skepticism about the announcement.
Without those clinical details, outside researchers cannot independently assess the company’s account of the treatment.
AI-Designed Viruses Already Replicated in Laboratory
The announcement follows earlier experiments by scientists at Stanford University and the nonprofit Arc Institute in Palo Alto, California.
As Slay News reported last year, those researchers used artificial intelligence to design complete genomes for bacteriophages, then chemically printed hundreds of proposed DNA sequences.
When tested with E. coli, some of the synthetic designs produced functioning viruses that replicated and destroyed the bacteria.
Of 302 designs, 16 successfully produced working viruses.
“This was a pretty striking, just actually seeing, like, this AI-generated sphere,” said Brian Hie, head of the Arc Institute lab.
That work involved bacteria in a laboratory, rather than treatment of human patients. Nevertheless, it demonstrated that AI-generated genetic designs could produce viruses capable of replication.
Synthetic biology pioneer J. Craig Venter warned about applying such capabilities to dangerous pathogens.
“If someone did this with smallpox or anthrax, I would have grave concerns,” Venter said.
The distinction is critical: bacteria-targeting treatments have potential medical benefits, while similar advances in biological design raise serious questions about misuse.
Researcher Warns Safeguards Have a “Fundamental Flaw”
Separate findings involving Chinese developer Moonshot AI have intensified concerns about dangerous information available through advanced models.
Earlier this month, researcher Peter Garrigan discovered that the company’s Kimi model could be manipulated into supplying information involving biological weapons, nerve agents, malware, aviation attacks, and assassinations.
“What we found is quite damaging and worrying,” Garrigan told Fox News.
Moonshot AI launched an internal investigation and began communicating with Garrigan about his findings.
The problem extends beyond Chinese systems, according to the researcher.
“We’ve also seen these problems within the U.S. models as well,” Garrigan said.
“It’s a fundamental flaw in the technology,” he added.
WATCH:
Those findings concern the misuse of AI-generated information, rather than Radical Numerics’ medical experiment.
Together, however, the developments underscore the growing stakes as AI systems become more capable in biology.
Radical Numerics is now seeking to commercialize a treatment it says has already reached a human patient.
Its promised peer-reviewed findings, and the clinical evidence needed to evaluate that announcement, have yet to be published.
Be the first to comment