Carlos “Charlie” Flores, Col, USAFR, (Ret)
Human responsibility, surveillance, and the danger of power without accountability
Artificial intelligence can recognize faces, analyze documents, and calculate the savings from eliminating thousands of jobs. But recognizing a face is not knowing the life behind it. Predicting a layoff’s financial benefits is not bearing the consequences for a family or community.
AI does not establish social priorities in a vacuum. People decide what to build, what data to use, and which outcomes to reward. My concern is not only that machines may escape human control. It is that they may remain obedient to people whose incentives reward indifference.
Whose humanity counts?
The world already reveals what happens when power discounts human dignity. Resource extraction in parts of Africa has not always benefited nearby communities. Asian supply chains generate prosperity while sometimes exposing workers to exploitation. Across the Americas, histories of dispossession and the vulnerability of migrant labor reveal similar tensions. In Europe, legitimate immigration debates can coexist with hostility toward those deemed outsiders. These circumstances differ, but they raise a common question: whose humanity becomes negotiable when wealth or power is at stake?
Imagine an executive using AI to justify closing a factory. The forecast may be accurate, yet it cannot determine what weight to give displaced workers, local businesses, or a town’s future. Someone has defined success. If the only objective is maximizing shareholder returns, AI can efficiently magnify that choice.
When technology monitors its users
This problem predates AI. In 2005, Sony BMG distributed CDs containing undisclosed copy-protection software that raised privacy and security concerns; the Federal Trade Commission later brought charges.[1] In 2006, Microsoft’s Windows Genuine Advantage program drew criticism for contacting Microsoft daily during a pilot without adequately informing users.[2] These were distinct licensing controversies, not political surveillance campaigns. Yet they illustrate why legitimate commercial objectives do not justify hidden monitoring or unlimited control over users’ devices.
Now imagine AI granted access to emails, essays, searches, or private communications. It could classify political opinions, map associations, and identify critics of government policy or corporate practices. Not every AI service collects such data, and monitoring does not inevitably mean repression. But an institution intent on silencing dissent could use AI to scale surveillance and retaliation. A machine need not hate a dissenter to identify one; it needs access, instructions, and an operator prepared to act.
An algorithmic flag is not proof. A prediction is not a verdict. No one should lose liberty, livelihood, or the right to disagree merely because a system classifies lawful dissent as a threat.
Autonomy—and the abuse of control
In January 2024, AI entrepreneur Mustafa Suleyman asked Fareed Zakaria, “How do we stay in charge? That’s the key question.”[3] In his September 2026 CNN commentary, Zakaria warned that increasingly autonomous, humanlike AI could outpace our ability to monitor and control it. His principle was “no autonomy without accountability to humans.”[4] In the accompanying interview, Suleyman emphasized human-set goals and safeguards.[4]
Their concern is essential. Mine extends it: what if AI performs exactly as intended, but the intention itself is exploitative or repressive? A system that identifies peaceful dissidents may be technically successful and morally indefensible. Meeting specifications does not establish legitimacy. Human control cannot become an excuse for abusing power.
Eyes on AI—and its operators
Designers must build security, privacy, testing, and auditability into their systems. Organizations must justify their uses, protect sensitive information, document consequential decisions, and provide remedies. Legislatures should establish enforceable protections, while independent courts, auditors, journalists, and civil society scrutinize both corporate and government power. Government cannot be the sole overseer when government itself may misuse AI. Due process, limits on surveillance, and the right to challenge automated decisions are indispensable.
AI can improve medicine, education, and public services. It can also expose injustice. These possibilities make responsible governance urgent, not optional. Intelligence is not wisdom; optimization is not justice; technological capability is not moral authority.
AI is not without a soul; it has the soul of its designer. Not a literal soul, but the imprint of human values and priorities. That imprint also comes from those who buy, direct, and deploy the technology. AI need not feel greed to exploit or hatred to persecute. Responsibility must follow the full chain of human authority.
There are no eyes in AI—but there must be eyes on AI, especially on those controlling it. The danger is not only that machines might escape us. It is that they may remain perfectly under human control while giving unchecked power unprecedented reach.
The question is not whether AI will someday learn to think like a human. The question is whether the humans controlling it will remember how to act like one.
Notes and bibliography
[1] Federal Trade Commission. “Sony BMG Settles FTC Charges.” January 30, 2007.
https://www.ftc.gov/news-events/news/press-releases/2007/01/sony-bmg-settles-ftc-charges
[2] Bangeman, Eric. “Microsoft Admits Windows Genuine Advantage Phones Home.” Ars Technica, June 8, 2006.
https://arstechnica.com/uncategorized/2006/06/7017-2/
[3] CNN. “Fareed Zakaria GPS,” interview with Mustafa Suleyman, January 7, 2024. Transcript.
https://transcripts.cnn.com/show/fzgps/date/2024-01-07/segment/01
[4] CNN. “Fareed Zakaria GPS,” Fareed Zakaria commentary “AI Is Starting to Look More Disturbing Than Sci-Fi” and interview with Mustafa Suleyman, September 20, 2026. Rush transcript.
https://transcripts.cnn.com/show/fzgps/date/2026-09-20/segment/01
Authorial note: The social and political analysis, historical analogy, and proposed governance principles are the essay’s arguments, not quotations or positions attributed to Suleyman or Zakaria. The CNN September 2026 transcript identifies itself as a rush transcript.
About the Author
Carlos A. “Charlie” Flores, Colonel, USAF Reserve (Ret.), is a retired U.S. Air Force Reserve officer and cybersecurity and systems engineering professional with more than four decades of experience spanning military communications, intelligence, information operations, cyber warfare, and defense systems engineering. His military career included command of the 710th Information Operations Flight, assignments supporting the Defense Intelligence Agency, and service as Air Reserve Component Advisor to U.S. Air Forces Central Command. In the defense industry, he held engineering and cybersecurity positions with Science Applications International Corporation, Lockheed Martin, and Northrop Grumman, contributing to the security of advanced weapons systems, including the F-22 Raptor, F-35 Lightning II, and LGM-35A Sentinel. He holds a Bachelor of Science in Mathematics from Inter-American University of Puerto Rico and a Master of Science in Systems Engineering from Southern Methodist University. His professional background informs his interest in technological accountability, cybersecurity, and the societal implications of artificial intelligence.