In a rare show of alignment, CEOs from OpenAI, Google DeepMind, and Anthropic—along with 85 experts across tech, biology, and national security—signed an open letter urging mandatory gene synthesis screening and recordkeeping, warning that AI could help people
The day the open letter went live, Dario Amodei, Demis Hassabis and Sam Altman were already facing a familiar problem: how fast to advance AI, how to regulate it, and how to prepare society for systems that grow smarter than humans.
Their difference—on pace. on oversight. on readiness—didn’t prevent them from signing a document about something far more specific than model performance. Along with 85 other experts in tech. biology and national security policy. the CEOs backed more robust rules for gene synthesis—an essential lab technology that can also be misused.
Gene synthesis is how custom DNA sequences are chemically built in a lab. It underpins modern microbiology and medicine: researchers can order synthetic genes from commercial DNA providers to develop new vaccines. drugs. and gene therapies for inherited diseases like hemophilia. produce human insulin. boost agricultural output. and more. It is also foundational to successful CAR-T cell therapies for cancer and used in many diagnostic tools.
The demand for synthetic DNA is growing globally, and it’s “never been cheaper or simpler to write genetic code.”
But the same tool that helps build life-saving therapies can assist in assembling deadly pathogens by arranging nucleotides—the genetic building blocks of life—in a dangerous order instead of one that matches natural templates.
Most US gene synthesis providers voluntarily screen orders for sequences of concern—such as those that can make a pathogen more dangerous or transmissible—and verify customers are legitimate. The safeguards exist because providers recognize the stakes, even when the rules are not legally required.
“It’s a real risk that bad actors could find a gene synthesis company with more lax standards, and that might mean disaster,” said Becky Mackelprang, director for security programs at the Engineering Biology Research Consortium.
Mackelprang’s concern is rooted in a simple reality: as long as screening remains voluntary, not every provider will screen.
James Diggans, vice president of policy and biosecurity at Twist Bioscience, offered a counterpoint. “This technology has been commercially deployed for more than 20 years and has never been misused to cause harm.”
AI changes the calculations. The open letter and its signatories worry that both large language models and AI biodesign tools can help design “entirely novel genetic sequences.” That matters because current screening systems often rely on similarity to known pathogenic or toxic sequences.
A system designed to catch an order for a known dangerous virus like Ebola might miss a new sequence that is still risky.
A study published in Science last year found that screening systems have kept pace with AI capabilities so far. Diggans said the industry understands this will not be the case forever.
Mackelprang isn’t arguing that AI alone turns a person into a bioweapons expert. She draws a line between knowing and doing.
“Researchers spend years trying to make a protocol work even after consulting directly with others who have perfected that exact same protocol. ” she told MISRYOUM USA News via email. “I think AI can help someone ‘level up’ their laboratory skills. but I do not think AI can enable someone without any biological training to create a serious hazard.”.
Still, she argues the near-term bottleneck matters—because gene synthesis companies are where anyone trying to produce a novel genetic sequence must pass.
“In the near term, I think the likelihood of these types of misuse are quite low,” Mackelprang said. “But when the potential consequences are severe and technologies continue to develop rapidly, we have a responsibility…to develop reasonable prevention and mitigation options.”
That is why Diggans, Mackelprang, Altman, Hassabis, Amodei, and other signatories from gene synthesis providers, tech and life science companies, entrepreneurs, and national security experts asked for a specific policy shift: mandatory gene synthesis screening and recordkeeping of orders.
The open letter is co-organized by the Institute for Progress and the Foundation for American Innovation.
It also calls for providers to record synthesis orders and sequence data “to support biosecurity investigations,” so that if something evades initial screening, it can be traced back to its source. The letter says “awareness of traceability itself deters misuse.”
DJ Kleinbaum, co-founder of the automated lab startup Emerald Cloud Lab and one of the signatories, framed the requirement as practical rather than dramatic.
“Screening every DNA synthesis order before it’s manufactured is the kind of unglamorous, common-sense step that prevents a much bigger problem later,” Kleinbaum said.
The CEOs’ signatures—coming from companies that compete hard in the AI market—were treated by the signatories as unusually meaningful precisely because they share an understanding of catastrophe potential.
“Support for screening does not depend on any particular view of AI,” the letter reads. “This is a rare moment of agreement across stakeholders that are often at odds.”
It presses Congress to act now, pointing to bipartisan legislative efforts tied to the Biosecurity Modernization and Innovation Act. That bill would give the Department of Commerce a year to develop new gene synthesis screening rules.
The letter also suggests US states implement screening requirements based on federal and industry guidelines to create a unified national standard instead of a patchwork of inconsistent laws.
This isn’t about applying biosecurity regulations to AI companies themselves, which the letter notes would limit the number of tech signatories. The open letter instead focuses on screening in a place where human action can still intervene before any dangerous sequence is built.
Even so, the signatories describe an ecosystem already moving—some of it directly and some of it indirectly—toward defensive AI.
Anthropic is hiring a technical chemical, biological, radiological, and nuclear threat investigator for its threat intelligence team. In May. after launching GPT-Rosalind. a frontier model intended to accelerate life sciences research and drug discovery. OpenAI introduced Rosalind Biodefense. a program allowing trusted developers to use GPT-Rosalind to build biodefense tools.
On June 4—the day after the open letter went live—security specialists at OpenAI and Anthropic served as panelists at the Bipartisan Commission for Biodefense’s meeting on AI and biological threats.
Diggans argued that the best defense against misuse of AI models to design harmful pathogens is to use AI models as defense.
The idea is for gene synthesis companies to use defensive models to ensure orders for highly engineered sequences get the same scrutiny as orders for naturally occurring sequences.
David Haussler, scientific director of the UC Santa Cruz Genomics Institute and a signatory, described what he thinks would be required.
“[Gene synthesis] companies have to agree to have their order screened not just against a list of sequences but by an AI that people agree is smart enough to recognize and thwart an adversary who’s trying to build a deadly pathogen,” Haussler said.
The letter’s supporters also point to work being done beyond gene ordering itself.
Last year, MISRYOUM USA News reported that OpenAI provided $30 million in seed funding to the biodefense startup Valthos. Valthos develops frontier AI systems to detect biological threats and create medical countermeasures, and Kathleen McMahon, the company’s co-founder, signed the open letter.
In September 2025. the Coalition for Epidemic Preparedness Innovations (CEPI) and philanthropic nonprofit Sentinel Bio created the Pandemic Preparedness Engine AI platform. sometimes referred to simply as “the Engine.” It takes a biosecurity-by-design approach. including “protecting biosecurity-sensitive data needed to train the AI” and managing access to the Engine while monitoring how it is used.
Users would interact with the system using AI prompts, similar to how people use consumer platforms. User prompts would be monitored in real time by a specialized AI agent designed to assess misuse risk or attempts to “jailbreak” a large language model to generate prohibited content. such as the “recipe” for assembling a deadly virus.
Even then, the signatories warn, commercially available systems can overcorrect.
This week, Anthropic launched Claude Fable 5, a version of its restricted Mythos model aimed at safe public use. The model automatically stops using Fable if it detects requests involving cybersecurity. biology. chemistry. or distillation—attempts to extract Claude’s capabilities to train competing AI models—then routes those requests to a less powerful model.
Users have complained that discussing biology for legitimate purposes with Fable 5 can lead the model to refuse or default to less capable versions. The letter’s argument, as reflected in Haussler’s concerns, is that safety barriers should not accidentally reduce beneficial life sciences work.
“The major providers of LLMs are doing their best to prevent the models from answering questions that would enable somebody to do something dangerous. ” Haussler said. “[But] the end product of jailbreaking an LLM that’s capable of teaching you how to build a deadly virus is that you now have an LLM that’s capable of teaching anybody how to build a very dangerous virus. And we don’t want that to happen.”.
The signatories believe this is exactly the kind of moment to act before an escalation becomes unstoppable.
Richard Danzig, a natural security expert who served as the 71st Secretary of the Navy under former President Bill Clinton and a signatory to the letter, said mandatory synthesis screening is rare in its feasibility.
“Mandatory synthesis screening is that rare case where a threat is clearly visible and substantial prevention clearly achievable before any crisis has occurred,” Danzig said. “Opportunities to act in advance are unusual in this field. I think we should take this one.”
AI safety gene synthesis DNA screening biosecurity OpenAI Anthropic Google DeepMind Biosecurity Modernization and Innovation Act Department of Commerce CAR-T biodefense