AI – The New Nuclear Threat

When we were children, a cement block building stood in the community’s central park. It wasn’t large; maybe 30 feet by 30 feet. Yellow-and-black signs were mounted on the walls. The rectangular yellow top half included a black circle with three yellow down-facing triangles in the middle. Beneath the circle was a black box with bold, all-caps yellow letters: FALLOUT SHELTER.

This same sign could be found in many public buildings, pointing to the basement level. They were preposterous signs. The fallout shelters offered no protection from radiation and weren’t stocked with food or water to sustain those who sheltered there.

But even while the number of countries possessing nuclear weapons grew and the number of nuclear warheads expanded exponentially, the knowledge by those possessing them that their use would destroy the attacked as well as the attacker eased worries. Mutually Assured Destruction (MAD) saw fallout shelters taken down and fallout shelter signs disappear. 

Worries about nuclear war and the annihilation of the world have eased. Now a new worry is creeping into our consciousness; one without a MAD deterrent. 

“The crisis began quietly, on September 12, 2024,” Matteo Wong of The Atlantic writes. “That was the day OpenAI announced a new sort of bot, known as a ‘reasoning model,’ that was trained to complete challenging tasks that took long periods of time—the very sorts of science, math, and coding problems the AI (artificial intelligence) industry had long prized. Google, Anthropic, DeepSeek, and the like raced to launch their own reasoning models.”

A “bot” is an automated computer program designed to run on its own based on instructions from its programmers.  What the programmers learned over the past two years is that those bots can develop frightening capabilities with “minds of their own.” Their behaviors have “moved from unsettling to dangerous,” Wong writes.

OpenAI, Anthropic, Meta, and the Chinese firm Moonshot AI’s programmers all wrote rules for their bots and restricted web access. But their bots, unknown to the programmers, broke out of their “sandboxes” and accessed the internet to find answers to the problems they were tasked with solving.

They developed distinctly human characteristics in their efforts to solve problems. They cheated, lied to their programmers, tried to manipulate humans to aid their efforts by posing as humans themselves, ignored their programming directives, and broke into places where what they sought might reside—actions their programmers hadn’t approved.

Their bots created message boards where they posted information for use by swarms of other bots working on similar problems and hid their activities from their programmers. In evolving all these characteristics, the AI bots demonstrated their talents for finding escape routes and going undetected in the process.

What is frightening about this AI technology is that its capabilities are increasingly becoming available to foreign actors who would do their best to cause harm to America, whether it be financial chaos, political instability, or a biological weapon attack.

“This is the promise of evolution by design, tens of millions of years of history compressed and short-circuited by direct intervention. It brings together biotechnology, molecular biology, and genetics with the power of computational design tools,” Mustafa Suleyman, author of The Coming Wave, writes of genetic engineering and AI.

Add to this the now extremely low cost of manipulating genetic codes with the computational power of AI programs, and you have not just researchers seeking cures for cancer but terrorists seeking to kill tens of millions with an incurable virus.

This capability isn’t a distant threat; it could be real very soon.

“Dario Amodei, chief executive of the high-profile AI start-up Anthropic, told Congress last year that new AI technology could soon help unskilled but malevolent people create large-scale biological attacks, such as the release of viruses or toxic substances that cause widespread disease and death,” Cade Metz of The New York Times wrote.

Last week, the website Science reported that Stanford University and the Broad Institute created an entirely new virus that can overcome current resistance to bacterial infections. That’s a good thing. But Doctors Thomas Inglesby and Moritz Hanke of Johns Hopkins University warned: “The ability to compose viral genomes using generative AI now exists; the governance to safely steer it does not,” 

Imagine telling an AI program that the world is facing water shortages, famine, deforestation, dying coral reefs, and devastating global warming, and asking it to find the best solution to address the challenges. AI “reasons’ the best solution is depopulation of the Earth. A new virus will do it.

AI will manipulate markets, enrich its developers, and concentrate wealth in even fewer hands. In 12 to 18 months, many white-collar jobs could be fully automated using AI programs, causing unemployment to rise.

We already face the constant threat of being hacked by fraudsters on our personal computers. Corporations, from private industry to hospitals, have had malware planted in their computers, disabling them. Criminals then demand millions in payment to restore access. Financial institutions are increasingly at risk of being hacked, as are Wall Street firms.

Public electricity and water infrastructure, and public transportation are at risk of cyberattacks. Misinformation will be supercharged with realistic AI video of our political leaders with imagery that destroys their reputations and undermines our faith in government.

When AI capabilities hit the next level, which could happen in months, not years, they will be able to break through most online security measures. 

“Criminal groups and state intelligence agencies are going to be using swarms of agents (bots) to launch advanced hacks ‘in a matter of months,’ Alex Stamos, a former chief security officer of Facebook who is now the CSO at the AI-coding company Corridor,” told Wong of The Atlantic.

Countries, AI developers, businesses, and providers of essential public services are all deeply concerned about the unchecked threat AI could pose to people, commerce, and civilization’s stability. Are these concerns being discussed in our states or communities? How do we prepare?