Something Scary Is Happening In Biolabs - Annie Jacobsen - #1127

Last time it was nukes; today it’s germs—so which one actually keeps you up at night, nuclear war or biological war?

Both are terrifying in different ways: a nuclear blast ends you at once, while a bio attack can be slow, invisible, and make you wish it had been instant.

Is bio war ignored because it’s illegal on paper and invisible in practice—no mushroom clouds, no warning?

Exactly; treaties say no one has them, yet engineered agents exist, and a pathogen is a living munition that replicates once it’s loose.

Is this the most underrated existential risk we face?

The Pentagon’s spectrum puts bio close to nuclear in damage and far more likely to occur, especially with dual‑use research that blurs defense and offense.

What shocked you most comparing the two?

There’s no plan to save the public from a major nuclear strike, but there is a plan for a bio catastrophe—only it prioritizes keeping the government alive.

Walk us back to the origins of gene engineering.

In 1971 Paul Berg created recombinant DNA at Stanford, the foundation for today’s gene editing; CRISPR is a later generation of the same idea.

What’s the case for doing this work at all?

Like fire, it heals or harms; we tweak dangerous bugs to learn countermeasures, which is also how you learn to make them worse—hence the gain‑of‑function fight.

Did COVID make us safer or just expose the cracks?

It proved how unprepared we were and wrecked trust in science and public health; rebuilding that trust is now as vital as any vaccine.

Toby Ord’s chart puts engineered pandemics at roughly one in thirty this century, unaligned AI at one in ten, and man‑made risks overall at about one in six; it’s the black‑ball problem of tech progress.

I wrote one plausible bio‑war scenario using engineered pneumonic plague, which is near instant in the lungs and functionally one hundred percent fatal if not treated at once.

Who took bioweapons industrial?

The Soviets built a vast secret program after signing the treaty; a defector exposed it, and British and U.S. experts confirmed large‑scale production and delivery systems.

Does the U.S. have anything like that today?

Officially only defense, but many practitioners say gain‑of‑function often crosses into de facto weapon engineering.

There’s also the grim logic of preemption.

Nukes may be considered to sterilize germ labs because conventional strikes can aerosolize pathogens; the escalation risk is obvious.

People used to fling diseased bodies over walls; the idea isn’t new, just the precision.

Modern engineering plus airborne spread makes containment brutal; once it’s out, it reproduces on its own clock.

What defines a nightmare bio agent?

Airborne transmission and extreme virulence; pneumonic plague kills unless hit with antibiotics within roughly a day, and engineering can erase that cure window.

The Soviet goal?

Kill the people, keep the infrastructure, and break faith in medicine by outfoxing antibiotics.

How did we learn all this?

A senior scientist, Vladimir Pasechnik, defected; London debriefed him, the CIA verified with top biologists, and the scale became undeniable.

Defenses also multiply risk by multiplying labs.

After 9‑11 and the anthrax letters, BSL‑3 and BSL‑4 labs exploded in number, synthetic biology arrived, and AI now accelerates what’s possible.

Play out day one of your scenario.

A blast at Russia’s Vector lab near Novosibirsk releases engineered pneumonic plague; U.S. space and signals intel light up, but denials burn through the twelve to thirty‑six hour window where you can still stop a pandemic.

You’re watching sensors, considering preemption, and stuck in a wicked problem.

We sniff for unusual decon chemicals with MASINT while knowing a strike could trigger nuclear exchange, so every option creates new problems.

Step one is knowing what the bug is.

Yes; dose matters, incubation in my case is about one to three days, and it takes surprisingly few particles to start an infection.

Longer incubation spreads wider; why choose short?

Reports say the Soviets tried adding a euphoria gene so the newly infected mingle before collapsing; I chose it because that’s scientifically diabolical and horribly plausible.

How does your chain of transmission kick off?

Travelers seed it worldwide, a student brings it to Los Angeles, and homeless encampments—absent from hospital surveillance—turn into amplifiers.

Can speed save us?

Speed is everything; we can deploy CBRN teams, but initial orders protect critical infrastructure, and by day six disorder forces continuity‑of‑government plans.

Devolution means what in practice?

Under COGCON, preselected officials grab go‑bags and disperse to hardened sites so the state endures even if the public is on its own.

What about the national stockpile and civil unrest?

The Strategic National Stockpile holds vaccines and antibiotics with twelve‑hour push packs; once people learn that, in a heavily armed country, those warehouses become flashpoints.

AI and info hazards are part of this too.

I once coaxed an AI into naming a likely site that later got scrubbed, which shows tech is outrunning oversight—even builders are surprised by what their systems enable.

COVID didn’t tip us into chaos, but it showed how brittle things can be.

A roughly one percent fatality rate kept society intact; double or far higher would look very different, so leaders must admit uncertainty, rebuild trust, and stop stigmatizing open questions like lab‑leak debates.

Let’s land it there; this was gripping and sobering—everyone should read the book; bookstores have it, and we’ll see you next time, my beauties.