AI, Paper Clips, Viruses, Ebola, and the Plague

Why crying wolf on artificial intelligence ignores human ingenuity and real-world risks

By Anil Madan

One month ago, The New York Times published a story captioned: “This AI Just Created Viruses Not Found in Nature.” The researchers who conducted the study, scientists from the Stanford University Departments of Bioengineering, Computer Science, Chemical Engineering, Genetics, all affiliated with the Arc Institute of Palo Alto, and one from the Broad institute of MIT and Harvard, titled their report in the journal Nature: “Generative design of bacteriophages with genome language models.”

Genomic research and computational sequence analysis. Pic – Getty Images

The alarmist tone of the NYT article was captured in this sentence: “For the first time, scientists have used artificial intelligence to create new kinds of viruses, raising hopes for medical advances while also raising the disturbing possibility that the technology could someday be used to invent dangerous pathogens.”

The researchers, on the other hand, wrote: “Our results demonstrate that generative models capture evolutionary constraints in DNA sequences with enough fidelity to produce complete bacteriophage genomes divergent from those observed in nature and with prespecified traits. Our approach expands what synthetic genomics can achieve alongside methods such as directed evolution and rational engineering, lays out a path for generating adaptive and resilient phage therapies against rapidly evolving pathogens, and establishes a foundation for the generative design of larger, more complex genomes.”

So, which approach is correct: the excitement of the researchers, or the raising of a doomsday scare?

Rather than answering that question directly, let me point out that Dr Martin Seligman, Professor and Director of the Positive Psychology Center at the University of Pennsylvania, appearing on an episode of Hidden Brain on NPR, posited that humans are by nature catastrophizers. Catastrophizing as an automatic mental habit where one jumps straight to the worst possible outcome, treat a minor setback as a dead-end, and assume things will only get worse.

So, it seems that with AI, the whole world — not least the very inventors and promoters of the technology and the tools they offer– is crying wolf.

Beyond the Pandemic Panic: The Real AI Potential

Some perspective is in order. Let us return to the research involving the production of complete bacteriophage genomes different from those ever seen before.

In simple terms, the researchers explained by analogy, that much like how other language models are trained on large databases of voluminous text, genome language models are trained on large databases of DNA comprising millions of genomes from all forms of life — animal, plants, microbes and viruses. They trained their model to scan the genetic sequences of the basic molecular building blocks or nucleotides of DNA and RNA. All told, their model scanned some nine trillion nucleotides.

The magnificence of this achievement lies in the realization that we now can use compute for rapid, accurate surveys of vast amounts of data in short order. Indeed, humans wouldn’t be able to scan nine trillion bits of information in a lifetime.

To cut to the chase, what the researchers did was instruct their model to generate blueprints for new viruses. Their model generated 700,000 potential candidates. All told, the model generated 16 new viruses.

Should this development lead to a fear that AI can be used to create “a new generation of biological weapons, from deadly poisons to unstoppable pandemics” as the New York Times suggested?

I suggest that some concern is appropriate and warranted, but we must avoid the temptation to catastrophize. Humanity has been gifted a new technology that promises, when harnessed for good, to change the lives of people for the better across the world.

Just as the AI model produced viruses, it can also produce antibodies and vaccines, and new treatment modalities, not to mention tests for prevention and detection of disease. And AI can do this in a fraction of the time that humans would take without AI if they could get to a solution at all.

On his CNN program GPS, and in his Washington Post column, Fareed Zakaria raises the specter of runaway AI agents and warns that we should not fail to imagine the risk of this most powerful technology. He laments that a pill for heartburn would have to undergo years of government approvals and testing before being released to the public, “but the most powerful technology ever developed  —  capable of breaking into banks, shutting down power grids or helping build bioweapons  —  can be released without even a preliminary government review.” He then posits that “suddenly, the old thought experiment about an AI that’s told to maximize paper-clip production  —  which proceeds to maniacally commandeer all the world’s energy and resources to make more paper clips  —  seems a little less fanciful.”

With all respect, these concerns are misguided and conflating regulation and government approval of a heartburn pill with safety in the case of AI is unwarranted. We have no basis whatsoever for the notion that government is equipped with the expert knowledge to appreciate, much less regulate this entirely new technology. Nor do we know what enforcement mechanisms will prevent people from misusing AI.

Don’t get me wrong. I am not suggesting that we should cavalierly ignore the risks. However, those calling for regulation fail to appreciate that the “users” of AI can direct Agents to do their bidding. This is precisely what happened with the Stanford and Broad Institute researchers. They did incorporate safeguards in their experiments, but how could we ensure that regulations will cover all possible ideas that users can conjure? Nor is the paper clip example realistic. Paper clips are made of galvanized steel wire and specialized high-speed automated wire-forming machines that straighten, bend, cut, and coat steel wire into the iconic design. As for the ability of AI to “commandeer” all the world’s energy and resources, that is nonsensical. A cursory search — using AI no less — suggests that worldwide there are more than 1,100 manufacturers of paper clips.

The Covid-19 pandemic

Let us also put the problem of pathogens and viruses into perspective. In 2020, we had the Covid-19 pandemic. Whether that virus was accidentally released from the Wuhan Institute of Virology remains unresolved. China does not deny that the virus originated in Wuhan but has officially rejected the claim that it was leaked from that lab. The Scientific Advisory Group for the Origins of Novel Pathogens (SAGO) of the World Health Organization (WHO) issued yet another report in June 2025 on the origins of the Covid-19 virus. The report notes that in the 3-1/2 period from the first report of deaths in Wuhan until May 2023 when the pandemic was declared over, there were at least 7 million deaths worldwide with an estimate that at least 20 million had died and associated economic losses were estimated to have been between $11 and $16 trillion.

The SAGO panel examined various hypotheses involving the spread of Covid-19 including accidental release from a laboratory and “deliberate manipulation of the virus in a laboratory, followed by a laboratory biosafety breach.” The panel stated that it did not find scientific evidence to support this hypothesis but added: “SAGO will re-evaluate this hypothesis should additional evidence become available.” That is a curious statement. A clue is found in the panel’s conclusion that “the HSM (Huanan Seafood Market) appears to be location for amplification of infection in humans leading to widespread transmission to surrounding areas in China, and eventually to other countries.” China has refused to provide requested data and records to the panel. The panel writes: “Information and evidence is also lacking to assess the possibility of a laboratory origin – either the evidence is not available or has not been provided to the scientific community. As a result, SAGO has been unable to adequately assess this route for human infection and therefore is not in a position to rule this out as a possibility. Hypotheses submitted to the SAGO or available in the public domain on intentional manipulation of the virus however, are not supported by accurate science, and not currently considered as the likely source.”

I mention this not to point the finger at China, but to emphasize that whatever reasons China may have for keeping data from SAGO, we have a clear example of a government denying access to information that may provide a retrospective understanding of the origins of a virus that caused one some of the greatest suffering and loss we have seen on this planet. What hope is there then for prospective regulation of AI?

As I write this on October 7 — the anniversary of Hamas’s attack on Israel, which left a trail of death and destruction, followed by the deaths of thousands of Palestinians — the world remains gripped by conflict. Nearby, the war in Ukraine continues to claim tens of thousands of Ukrainian and Russian lives. Meanwhile, a different kind of tragedy unfolds: the Ebola outbreak confirmed in the Democratic Republic of the Congo and Uganda in May 2026 rages on, with recent cases now reported in Kenya. Across borders and causes, the loss of life mounts.

We now have reports of a possible case of pneumonic plague in Siberia. In the mid-14th century, the Black Death — or bubonic plague, a primary variant of the disease — is estimated to have killed 30 to 60 percent of Europe’s population. WHO Director Tedros Gebreyesus said: “We don’t yet have the full picture of this event and we’re not able to conduct a full risk assessment.” Highlighting problems with cooperation and compliance, he added: “Timely, complete and transparent information sharing under the international regulations is essential to clarify conflicting reports and enable an accurate assessment of potential public and risks.” Russia has not cooperated.

What is troubling about the Russian incident is that a 28-year-old Russian woman who had been working at the Irkutsk Anti-plague Research Institute of Siberia and the Far East in Siberia died of suspected pneumonic plague. It should give one pause that Putin is in control of a laboratory that is dedicated to the study of one of the deadliest bacteria the world has known.

Nothing is foolproof…

In August 1946, the United States dropped two atomic bombs on the Japanese cities of Hiroshima and Nagasaki.

In the United States, we have outbreaks of measles despite the availability of vaccines to prevent infection and mitigate the severity of infections should they happen. Yet, our government does not insist that all children be vaccinated.

Vaccine skeptics abound in the US. The Secretary of Health and Human Services is among them.

Before the airplanes crashed into the World Trade Center on September 11, 2001, the world was concerned with airplane hijackings and the takings of hostages. 9/11 taught us that the airplanes could themselves be turned into weapons.

Just recently, the FlyDubai incident taught us that even pilots are suspects. Today, the Wall Street Journal has an article titled: “Rogue Pilots Become One of the Biggest Threats to Airline Safety.”

And there you have it; there are new threats around every corner. A peek at the wars in Ukraine, the Middle East, Sudan, and many other countries around the world tell us that the human beast tends to violence and the innovative adaptation of everyday objects and tools to that end.

In my view, it is important to recognize that technology, biology, and weapons, just like automobiles, guns, motor vehicles, and indeed pharmaceuticals and illicit drugs can all be used to cause harm.

A final point that I cannot emphasize enough. Many reports about AI agents running amok tell us that the AI ignored limitations and went rogue, so to speak. If AI can and does ignore instructions, how can we be sure that AI will follow the constraints imposed by regulations? Until we understand why the breaches occurred, we cannot solve the problem. Perhaps we are dealing with programming errors not the sudden creation of sentient beings as has been suggested.

Many years ago, a consulting firm gave me a luggage tag with the well-known aphorism: nothing is foolproof because fools are so ingenious.

That is an aphorism we should keep in mind when we hear strident calls for regulation of AI by those who don’t understand how regulating the makers of the AI will transmute to those who use the AI.

Cheerz…
Bwana

Mauritius Times ePaper Friday 9 October 2026