AI can explain hard bi
ology in simple steps.
Synthetic DNA and RNA can be ordered from companies.
Lab tools are getting cheaper and easier to use.
This does not mean any person can create a biological weapon.
But it does mean the old safety walls are weaker.
On June 4, 2026, a public letter asked Congress to create stronger rules for synthetic DNA and RNA screening.
The letter was titled: “In Support of Mandatory Nucleic Acid Synthesis Screening and Recordkeeping.”
It was organized by the Institute for Progress and the Foundation for American Innovation and signed by major AI and technology leaders, including:
- Sam Altman — OpenAI
- Dario Amodei — Anthropic
- Demis Hassabis — Google DeepMind
- Mustafa Suleyman — Microsoft AI
- Alexandr Wang — Meta
- Patrick Collison — Stripe
- Paul Graham — Y Combinator
Scientists, national security experts, and synthetic biology company leaders also signed it.
The letter asks Congress to require companies to check synthetic DNA and RNA orders before making or shipping them.
The main protections are:
- Screen the genetic sequence
- Verify the customer
- Keep records of orders
Synthetic DNA is useful. It helps medical research, vaccine work, and disease study. It lets smaller labs do serious work.
That is the risk: AI can lower the knowledge barrier.
In 2009, the International Gene Synthesis Consortium, or IGSC, whose members represented about 80% of global commercial gene synthesis capacity. created voluntary safety practices.
Its Harmonized Screening Protocol v2.0 was dated November 19, 2017.
It says member companies should screen orders, check customers, review risky orders, and keep records.
If 80% is covered, then 20% may not be.
On April 29, 2024, the White House Office of Science and Technology Policy released a Framework for Nucleic Acid Synthesis Screening.
Later, the National Institutes of Health said NIH funds may only be used to buy synthetic nucleic acids or benchtop synthesis equipment from providers that follow the framework.
In February 2026, Senators Tom Cotton and Amy Klobuchar introduced bipartisan legislation to screen sales of potentially dangerous synthetic DNA.
The proposal would require screening rules for labs and customers.
It would also ask the Commerce Department to help identify dangerous genetic sequences.
A 2025 Science study, described by the International Biosecurity and Biosafety Initiative for Science, tested whether AI could redesign harmful proteins.
Researchers created 76,089 variants of 72 proteins of concern, mostly known toxins.
At first, screening tools did not always catch them. Then the researchers worked with screening developers and DNA synthesis providers, including Twist Bioscience and IDT, to improve detection before public release.
This shows the danger clearly. AI can change a harmful sequence so it no longer looks like the original. If screening is too simple, it may miss the risk.
The National Institute of Standards and Technology, or NIST, has warned that AI may help design new DNA sequences that current screening tools may not detect.
Its project page was created on March 25, 2024, and updated on May 22, 2026.
This shows the problem is real enough for government action.
A 2025 study by researchers from the Center for AI Safety, MIT Media Lab, UFABC, and SecureBio tested AI models against PhD-level virologists on hard lab troubleshooting questions.
The study reported:
- PhD-level virologists scored 22.1% on average in their own areas
- OpenAI’s o3 scored 43.8%
- Google’s Gemini 2.5 Pro scored 37.6%
This does not mean AI can run a lab safely.
It does not mean AI can create a weapon alone.
But it means AI can give serious technical help.
The law should protect real research. But it should not allow anonymous or suspicious orders of dangerous genetic material.
AI companies also have responsibility.
Screening tools must also improve.
The answer is not to stop science. The answer is clear rules:
Screen the order.
Verify the customer.
Keep records.
Update screening tools.
Protect real research.
Block dangerous misuse.