Annie Jacobsen, author of Biological War: A Scenario, told NewsNation she believes current AI development resembles the early, tightly controlled era of thermonuclear weapons. She warned that mixing AI with weapons of mass destruction is especially dangerous and that government-level controls are largely absent today. Jacobsen says she demonstrated a ChatGPT exploit that revealed the Strategic National Stockpile's location in Los Angeles and that the behavior was later removed after she raised it with Sam Altman. She calls for stronger oversight and broader expert engagement.
Annie Jacobsen Warns: Today's AI Echoes the Thermonuclear Era — Risks of Mixing AI With WMDs
AI compared to thermonuclear weapon: Annie Jacobsen
Annie Jacobsen, investigative journalist and author of the new book Biological War: A Scenario, warned on NewsNation's On Balance that the current stage of artificial intelligence development bears troubling similarities to the early days of thermonuclear weapons research.
Jacobsen, a Pulitzer Prize finalist, says the combination of powerful AI tools and weapons of mass destruction poses unique dangers because today’s development lacks the strict government control that governed thermonuclear programs in the 1950s. She told NewsNation she has discussed these risks with subject-matter experts around the world.
Demonstration and Dialogue With Industry Leaders
"There we are in this moment in time and why AI is so dangerous when you're mixing it with weapons of mass destruction, it's analogous to, say, 1951, 1952, when government scientists were developing the thermonuclear weapon," Jacobsen said.
Jacobsen also described a direct exchange with OpenAI CEO Sam Altman. She said she was able to prompt ChatGPT to produce information that—according to her account—revealed the location of the Strategic National Stockpile in Los Angeles. Jacobsen says she raised the issue with Altman and that the specific ChatGPT behavior she had exploited was later removed. She emphasized, however, that the incident underscores how many vulnerabilities may remain unknown.
What This Means
Jacobsen urged stronger, government-level oversight and broader international engagement to address AI risks, especially where AI could interact with biological or other weapons of mass destruction. While the claims she reports come from her investigations and conversations, they highlight calls from some experts for clearer safeguards, transparency, and independent review of advanced AI systems.
For readers: Jacobsen's account of the ChatGPT demonstration is reported as her claim; OpenAI has not independently confirmed the exchange described here in this article.
Help us improve.
