Tim van Opijnen | Scientific risk only matters when it's politically convenient
Scientific risk is taken seriously in the USA only when it is politically convenient, writes Tim van Opijnen. As a result, his own research is now coming under increased scrutiny. “We have to consider removing those experiments. Stopping the work. Changing our language. We are effectively being subjected to a ban.”
Twenty years ago, finishing my PhD and looking for what came next, I visited a lab at the top of a skyscraper right next to Central Park. They had just re-engineered the influenza virus responsible for the 1918 pandemic, done in a maximum-security facility where you work in a kind of space suit. I chose Boston instead, but their science went on to illuminate what made that virus so lethal, and what we might do to prevent the next pandemic.
A few weeks ago, I was on a call with a program officer at the National Institutes of Health, the largest funder of life sciences research in the world. The message came politely, as these things do. The experiments at the heart of one of our applications fall under gain-of-function research, that is: experiments in which pathogens are forced to change so we can understand how they might become more dangerous in the future. GoF, as it's known, is no longer something NIH can support. We should consider removing those experiments from our proposal. Stopping the work. Changing our language. We are effectively being handed a ban.
Our GoF work involves letting bacteria mutate and evolve in a controlled biosafety environment, observing how they overcome antibiotics and immune attack simultaneously. Every experiment requires my sign-off, institutional review, federal approval, and oversight from multiple external bodies. We do not do this lightly. This work produced a discovery we are publishing in one of the highest-impact journals in the world: a new mechanism by which bacteria evade both drugs and the immune system at once, usable as a diagnostic marker and the basis for an entirely new class of treatments. Had we started this work today instead of four years ago, we would have been stopped before we began.
Flying blind
Gain-of-function research is politically radioactive for one reason: COVID-19. The theory that the coronavirus originated at the Wuhan Institute of Virology, possibly through GoF experiments, is not without some scientific basis, though most scientists consider it unlikely. Either way, we are shutting down the most carefully regulated science in the country because it has been made politically toxic. Future outbreaks will find us less prepared. We are not learning the lessons of COVID. The fire station is being burned down. Methodically, and with a clear conscience.
Best baby
GoF research demands strict regulation. But this ban has more to do with political optics than biosafety. The rules apparently apply most forcefully where the money doesn't. That is how Nucleus Genomics came to plaster an advertisement across the New York City subway in November: “Have your best baby”. The promise: taller, smarter, healthier children, selected from IVF embryos. Your eugenics-alarm should be tingling. Last week, a team from Columbia University published research showing they had edited the DNA of human embryos with unprecedented precision using the newest tools called base editing. The lead researcher called for a public conversation about the implications. The financial potential is enormous, and because the federal government does not fund embryo research, Nucleus Genomics has opportunistically stepped into that gap, funding the next stage of Columbia's work. He Jiankui, the Chinese scientist who attempted something similar in 2018, served three years in a Chinese prison for it. His successors now have a marketing budget.
Embryo editing does not fall under GoF, but shares the same regulatory vacuum: no federal mandate, no democratic debate about whether this is what we want. Unlike GoF research into infectious diseases, this new form of eugenics carries a financial potential that drowns out all restraint. The field has even borrowed its name from a dystopia: Silicon Valley investors call their vision ‘the Gattaca stack’, after the 1997 film about a world stratified by genetic advantage. They have optimized the dystopia into a business model.
The exception
GoF research has been made politically toxic, not because of its risk, but because of the political utility of that risk. The same financial motive that frees embryo editing also protects artificial intelligence: private capital, no federal mandate, no democratic debate, at a scale that makes embryo editing look modest. Without regulation, reputation and conscience are the only possible brake on unwanted innovation. Last year, the AI company Artisan plastered ads across that same subway urging companies to 'Stop Hiring Humans.' The backlash made them apologize. Bad press was the only brake from outside.
Sometimes that brake comes from within. In April, Anthropic announced an AI model called Claude Mythos Preview, so capable at finding and exploiting software vulnerabilities that the company chose not to release it publicly. Now, two months later, Anthropic calls for a coordinated global pause on advanced AI development, warning that the models are showing signs of escaping human control. But this only works if every major AI company stops at the same time. In other words: one company with a conscience solves nothing. Every other AI company can simply keep going. There is no law that would stop them.
The choice not to regulate is itself a regulation, one that always favors whoever is moving fastest. My grant application is being rewritten as I type this, experiments removed, language adjusted to what the government will fund. The US economy has gone all-in on AI; if the bet doesn't pay off, the correction will make the financial crisis of 2008 look like a rough quarter. That is why nobody in Washington is writing the law that would require every company to do what Anthropic does voluntarily. So instead we cancel the GoF grants, bury our heads in the sand on infectious diseases, and let the subway sell you your best baby and your AI replacement. We can all see the cliff we are running towards, hoping there is no abyss.
