iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Eric Nguyen argues that AI developers should build biological safeguards alongside models that can design biological sequences. In a September 23, 2026 episode of Latent Space: The AI Engineer Podcast, the Radical Numerics co-founder and CEO describes this as a “dual mandate”: design capabilities may also help assess sequence properties, but that possibility is not proof that current models can reliably screen for biological risks on their own.
What Nguyen means by an AI biosecurity arms race
Nguyen’s argument is that biological sequence design and defense are connected capabilities. A genomic language model—a large language model trained on DNA—can be used to read sequences and predict properties, as well as generate new sequences. In his view, improvements in modeling could therefore support both design and efforts to discriminate or predict sequence properties.
That is the opportunity behind his “arms race” framing, not a claim that a measured contest between attackers and defenders has been established. The episode presents the prospect of defense lagging behind design as Nguyen’s concern. It does not provide independent measurements of that gap or establish that sequence-prediction models are reliable stand-alone biosecurity systems.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallCrashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteWhy Nguyen calls for a “dual mandate”
Nguyen says teams developing models that can design function into biological sequences should also take responsibility for safeguards around those capabilities. The logic is that the same broad modeling advances that make sequence generation more useful may help researchers evaluate sequences, so defensive work should be developed alongside design rather than treated as an afterthought.
#1 Best Overall
His conclusion is captured in the episode transcript: “The design side is going to get more capable. The defensive side needs to try to get ahead.” This is Nguyen’s position, not a regulator’s assessment or a consensus finding attributed to an independent expert body.
What the RNA aptamer example shows—and what it does not
The episode page describes an RNA aptamer experiment involving a large dataset. Nguyen says the model was shown lower-scoring examples in increasing-score order, while some top performers were held out. Asked to continue the pattern, it later recapitulated some of those unseen higher-scoring examples.
This is a reported example of a model identifying a pattern in a particular setup. The episode page does not supply enough experimental detail to establish how broadly the result applies, and it should not be treated as proof of general biological reasoning, autonomous discovery, or biosecurity effectiveness. It also does not show that a model can replace other screening or review methods.
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Repair Windows errors before they cause bigger problemsFix Now →Who Nguyen is and what the episode covers
Latent Space’s episode page identifies Eric Nguyen as co-founder and CEO of Radical Numerics, and says he previously helped develop genomic language models including Evo and Evo 2. The interview focuses on how sequence-design capabilities might relate to biological defense and why Nguyen believes safeguards deserve attention as those capabilities advance.
The episode page also summarizes work in which models were used by a separate Arc Institute and Stanford team to generate bacteriophage genomes that were synthesized into functional viruses. That is a consequential claim, but an episode summary is not a substitute for the underlying scientific record. The page links to a bioRxiv paper; detailed claims about that work should be grounded in that original paper rather than inferred from the interview summary.
Quick Recap
Best Value
Rank #4
How to interpret the episode’s claims
- Attribute the arms-race framing: Nguyen presents it as his view of how design and defense may develop, not as a settled scientific consensus.
- Separate potential from demonstrated reliability: the episode describes a possible defensive use for sequence models but does not establish them as dependable stand-alone screening systems.
- Keep the aptamer result in context: it is a specific reported example, with experimental details insufficient on the episode page to support broad generalization.
- Check primary evidence for consequential results: for the bacteriophage work, consult the linked original scientific paper for methods and findings.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
