The debate over whether artificial intelligence could pose an existential threat to humanity may sound unprecedented. But scientists have confronted a remarkably similar dilemma before.
More than half a century ago, researchers working with recombinant DNA realised that a technology capable of transforming medicine could also create biological risks that they did not yet know how to control.
Their response was to stop, debate the risks and establish rules.
The result was not the collapse of the technology. Instead, the framework eventually helped create the modern biotechnology industry.
That history is now being revisited as AI developers, scientists and policymakers struggle with a similar question: how can a potentially transformative technology be allowed to advance without leaving society exposed to risks that its creators cannot fully predict?
The Asilomar experiment
In 1975, around 150 scientists, lawyers and other experts gathered at the Asilomar Conference Center on California’s Monterey Peninsula to discuss recombinant DNA.
The technology was still only a few years old. Scientists had learned how to combine genetic material from different organisms, opening the door to new medicines and cheaper production of existing drugs such as insulin and antibiotics.
But the same techniques could also produce organisms with dangerous characteristics.
Some of the earliest experiments involved DNA from cancer-causing viruses, while others produced organisms resistant to antibiotics. Scientists began warning that the technology could potentially create biological threats on a scale that had never been seen before.
In two letters published in 1973 and 1974, researchers called for caution and eventually urged a temporary moratorium on recombinant DNA experiments until appropriate safety measures could be established.
The work was paused while scientists prepared for the Asilomar meeting.
The organisers faced a difficult choice. They wanted scientists to establish their own rules rather than leave the matter entirely to Congress, where lawmakers might impose regulations that were either too restrictive or poorly adapted to rapidly changing science.
After four days of intense debate, the participants agreed on a system of physical containment based on the potential risks of different experiments.
The National Institutes of Health subsequently adopted and enforced the framework.
The strictest laboratories required measures such as negative-airflow rooms and pressurised protective suits. Certain organisms were designed to die outside controlled laboratory environments, while experiments involving deadly toxins or antibiotic-resistant bacteria were prohibited.
When self-regulation was not enough
The Asilomar agreement did not end the controversy.
Public concern grew, particularly after local officials began questioning whether the new experiments could accidentally release dangerous organisms.
The most significant confrontation came in Cambridge, Massachusetts, when Harvard University sought permission to modify a laboratory for high-level recombinant DNA research.
Mayor Alfred Vellucci threatened to impose a citywide ban and demanded that scientists explain the technology in language ordinary residents could understand.
“We don’t understand your alphabet,” he told them.
Cambridge eventually imposed a three-month moratorium and created a citizen board to assess the risks. The board included people from outside the scientific community and visited laboratories at the Massachusetts Institute of Technology to examine the safety measures adopted after Asilomar.
The city also organised a public hearing that effectively put recombinant DNA on trial, with scientists arguing on opposing sides.
The result was the first municipal biosafety ordinance in the United States. It broadly followed the Asilomar framework but also imposed additional restrictions and, crucially, gave members of the public a direct role in deciding how the technology should be regulated.
The episode demonstrated that scientific self-regulation could provide a starting point, but public oversight and enforceable rules were also necessary.
Regulation did not stop biotechnology
One of the most important lessons from the episode is what happened next.
The regulations did not prevent biotechnology from developing. In fact, the industry began expanding rapidly after the Asilomar era.
Cambridge itself eventually became one of the world’s leading biotechnology centres. The regulatory framework provided companies and researchers with clearer rules about what was permitted and what safety standards were expected.
By 1982, even Vellucci had changed his position. At the opening of a Biogen facility, he reportedly said he had no objection to recombinant DNA as long as the company paid its taxes.
The economic success of biotechnology has become an important part of today’s debate over AI regulation. Supporters of stronger controls face an argument that tighter rules could slow technological development and allow other countries to gain an advantage.
The history of biotechnology provides a more complicated picture: regulation can also create certainty and public confidence, allowing an industry to grow within clearly defined boundaries.
The AI comparison
The parallels with artificial intelligence are increasingly difficult to ignore.
AI companies now face concerns about job displacement, misinformation, privacy and the possibility that increasingly autonomous systems could contribute to cyberattacks or biological threats. Data centres are also generating disputes over their consumption of electricity and water.
AI leaders themselves have repeatedly warned about the technology’s potential risks.
In 2023, Elon Musk, Apple co-founder Steve Wozniak and others called for a temporary pause on the development of systems more powerful than GPT-4. Two months later, leaders from OpenAI, Google DeepMind and Anthropic warned that AI could pose an existential threat and argued that managing the risk should be treated with a level of seriousness comparable to pandemics and nuclear weapons.
Yet the development of increasingly powerful AI systems continued.
That has fuelled scepticism about the industry’s calls for regulation, with some critics arguing that companies have become more vocal about catastrophic risks as they face increasing legal and financial exposure.
There is also a fundamental difference between the two technologies.
Recombinant DNA could be physically contained. AI cannot simply be placed behind sealed doors and protective equipment. Its risks can spread through software, networks and information systems across national borders.
That makes the question of enforcement considerably more complicated.
Who is accountable when AI goes wrong?
The experience of recombinant DNA suggests that establishing rules is only part of the challenge.
The more difficult question may be who has the authority to enforce them, who is responsible when an AI system causes harm and who has the right to investigate what happened.
Luis Campos, a historian of science at Rice University who has studied the recombinant DNA controversy, argues that the crucial lesson of Asilomar was not simply that scientists agreed to regulate themselves.
It was that they accepted that scientists could not be the only people making decisions about technologies with consequences for society as a whole.
That may be the most important lesson for AI.
The Asilomar model was never simply about taking scientists to a beautiful location in California and asking them to decide the future of technology. Its lasting impact came from the combination of scientific expertise, legal scrutiny, public involvement and eventually enforceable rules.
AI presents a much harder regulatory challenge because its risks cannot be physically contained.
But the history of recombinant DNA suggests that the choice may not have to be between unrestricted innovation and shutting down technological progress.
A third option exists: establish rules that make innovation safer, more predictable and accountable before the consequences become impossible to control.
Be First to Comment