Model poisoning and geopolitical risks

BONUS: Saving Democracy—How AI Is Transforming the Battlefield for Our Minds

In this very special BONUS episode, we speak with Anthony Vinci, former CTO and Associate Director of the National Geospatial-Intelligence Agency (NGA) and author of The Fourth Intelligence Revolution. Anthony has been at the frontlines of modernizing the intelligence community for the age of AI, and in this episode, he lays out a stark warning: we are entering an era where machines don't just augment intelligence—they transform it. But the real battlefield isn't just digital; it's cognitive, economic, and societal.

From Startup Founder to Intelligence Modernizer

"When I started my career, it was kind of the last dot-com boom... then I went into intelligence and became a case officer who goes out and recruits sources. I went to Iraq and places like this."

Anthony's career has uniquely zigzagged between the tech industry and the intelligence community. Starting in a New York startup during the 2000 dot-com era, he later became a case officer before returning to the startup world. When NGA needed someone to bring AI and modern technology into the agency, Anthony's rare combination of intelligence experience and tech entrepreneurship made him the ideal candidate. At NGA, he led the effort to implement computer vision and machine learning into workflows that were historically manual—where analysts would literally print satellite imagery and examine it with magnifying glasses. Nine years later, NGA now produces intelligence reports with "no human hands" involved.

The Automation Arms Race

"I believe where we're entering now is where the machine, the AI, has to do the analysis itself. Period. And it never comes to a person."

The volume of data has surpassed what humans can process, regardless of how sophisticated our tools become. Anthony points to a recent Anthropic report showing Chinese actors used Claude to automate 80-90% of a cyber espionage campaign. He believes we're approaching a world where 100% of cyber operations—both offensive and defensive—will be automated. The parallel he draws is striking: just as quantitative hedge funds trade in microseconds without human intervention because competitors do the same, cyber warfare and eventually physical drone warfare will follow this pattern. The only way to defend against automated attacks is to automate your defense.

How Social Media Already Threatens Democracy

"The longer a user was on TikTok, the more they used it, the more benevolent view of human rights in China that user had. So it's actually working, and it's so subtle, you can't even see it unless you do these big statistical studies."

The threat isn't theoretical—it's measurable. Researchers at Rutgers demonstrated that TikTok doesn't just censor content about the Uyghurs or Tiananmen Square; prolonged use of the platform actually shifts users' views on Chinese human rights. And that's just one piece of evidence, there are more! Unlike the 2016 election interference where the Russian Internet Research Agency placed targeted ads, modern influence operations work through algorithmic content selection. The platform doesn't need to show you propaganda; it simply needs to decide what you don't see.

AI Will Hack Our Minds

"AI is a dialogue. AI becomes this arbiter of information... This is really, really different when it comes to information operations. It's more like what I used to do as a case officer, where I'm trying to convince you of something."

Recent studies in Science and Nature demonstrate that AI systems trained for political persuasion are dramatically more effective than traditional advertising—not through persuasive rhetoric, but by overwhelming users with an abundance of "facts" (which aren't always factual). Anthony warns that the 2026 and 2028 elections will see widespread use of these tools. More alarming: Anthropic research shows that just 250 documents can poison a large language model. Foreign adversaries don't need millions of data points to corrupt the AI systems we increasingly rely on for information.

The Fourth Intelligence Revolution: What Must Change

"The first thing that we need to do is to compete in intelligence in those fields as well... economics, science, technology. And doing that requires intelligence to work with private companies, with the public."

Anthony outlines a three-part solution:

• Expand intelligence scope: Move beyond traditional political and military focus to include economic, scientific, and technological competition with China and other adversaries through a whole-of-society approach

• Automate everything: Embrace AI across all intelligence functions—it's the only way to compete against adversaries who are already automating

• Democratize resilience: Since everyone is now a target of foreign information operations, we can't rely solely on government protection. Citizens must learn to think like intelligence officers

Think Like an Intelligence Officer

"No matter how trusted the source, they're always going to look at another source. If you read the New York Times, go read Newsmax, or vice versa. And if they both say the same thing, that probably means it's true, or more true."

Anthony offers practical advice for personal information resilience. First, acknowledge you are personally being targeted—this isn't paranoia, it's the new reality. Second, triangulate information like an analyst: never trust a single source, and deliberately seek out opposing viewpoints. Third, think like a technology officer: before adopting any new app or platform, research who made it and assess the risks. This doesn't mean avoiding risky technologies entirely—it means using them with awareness and mitigation strategies like VPNs, limiting shared information, or using multiple accounts.

Name the Threat

"One thing is to think about the threat and to think that there may be someone who's targeting you... not just generally—me as an individual."

The core message is clear: the threat to democracy is the capability of adversaries to influence our views to go against our own interests. Whether it's voting behavior, economic decisions, or social cohesion, foreign actors now have the tools to target individuals at scale with personalized influence campaigns. The first step in defense is naming this threat openly. The book The Fourth Intelligence Revolution provides both the warning and a framework for response.

About Anthony Vinci

Anthony Vinci is the former CTO and Associate Director of the National Geospatial-Intelligence Agency (NGA) in the USA, and author of The Fourth Intelligence Revolution. He has flip-flopped between the tech industry and intelligence throughout his career—starting in a New York startup during the dot-com boom, becoming a case officer who served in Iraq, founding and exiting a tech startup, and then returning to government to modernize NGA for the age of AI. He is now CEO of Vico, a startup building AI for intelligence analysis.

You can link with Anthony Vinci on his website and subscribe to his Substack, 3 Kinds of Intelligence.