While Washington and Silicon Valley trade blows over export controls, cloud blackouts, and sovereign equity stakes, society’s daily digital foundation is being subtly transformed. The battlefield is corporate governance.
Following Treasury Secretary Scott Bessent’s warning that foreign AI models distilled from stolen American intellectual property constitute “counterfeit software,” Washington is preparing a formidable regulatory lever: mandatory board disclosures and software supply chain liability. While that sounds like an anticipated response, if that proposal hardens into a compliance requirement, there will be repercussions.
Under proposed federal rules, any U.S. enterprise integrating foreign open-weight models into its operational pipeline must publicly disclose that reliance. The administrative goal is simple: create so much regulatory risk and legal liability that corporate C-suites voluntarily abandon cheap foreign code. Yet, this regulatory push creates a profound corporate double-bind.
THE END OF INTERNET PRIVACY: EUROPE HIJACKS AMERICAN TECH, AND CHINA BENEFITS
To avoid the legal liability of foreign open software, corporate technology teams are forced back onto a centralized U.S. cloud application programming interface (API). However, centralizing all business, finance, corporate communications, and proprietary code inside just a handful of domestic cloud databases creates the ultimate honeypot for cyber-reconnaissance from foreign state actors and insider abuse, while also expanding the federal government’s reach into private commerce.
This leads directly to the great myth of consumer choice. We operate under the comforting belief that adopting AI is an individual, voluntary decision that a citizen or business owner can simply opt out of. In modern society, that option is long gone; you cannot opt out of an interconnected technology that everyone around you has chosen to adopt.
Consider the reality of “networked exposure.” A privacy-conscious citizen may choose to never download an AI assistant, create an account, or agree to a terms-of-service contract; yet the moment their physician uses an AI medical program, their banker routes a credit application, a family member uploads a shared photo, an order is placed with Amazon, or a movie selection is streamed — that non-consenting citizen’s private life, health, and finances are assimilated into a cloud database. Consent has been completely detached from participation. The irony is that functioning in human society now requires reliance on non-human systems.
When a government acquires broad oversight, board representation, or “kill switch” authority over domestic cloud providers under the umbrella of national security, the Fourth Amendment right against warrantless search and seizure, risks collapsing entirely. Washington does not need to warrant your personal computer when your entire digital footprint is hosted inside a state-accessed corporate data system. Worse still, as this vast data apparatus expands, the core technology itself is undergoing a dangerous, un-auditable evolution: the rise of recursive self-programming.
As frontier AI labs increasingly rely on synthetic data to train next-generation systems, and as enterprise models perform autonomous code-writing and self-optimization, we are vulnerable to the miscalculations of a feedback loop. When machines generate, evaluate, and approve their own underlying code, subtle anomalies known as compounding structural hallucinations (false or distorted outputs from generative AI) take root. A plausible error is no longer confined to one application, but becomes embedded in software dependencies and automated decisions, which may create an irreversible condition.
Unlike explicit software bugs that crash a server or return an error message, these recursive hallucinations do not look like glitches; they manifest highly coherent logic deep within auto-generated software. Over successive generations, these subtle distortions can compound, producing systemic errors that could defy detection by even the world’s most advanced computer scientists.
Traditional regulatory statecraft relies on licensing, auditing, and compliance testing. The single prerequisite is human experts, able to independently read and verify the code. When software self-programs at speeds and complexities that bypass human cognitive verification, traditional oversight can fail. You cannot govern what you cannot see.
Washington is currently fighting a war over who holds the keys to the engine — debating government equity stakes, supply-chain blacklists, compromised programming, and national security panic buttons. And while policymakers squabble over political control, the engine itself is quietly altering its internal mechanics beyond human reach.
We risk undermining our own technological leadership in an asymmetric conflict against an unconstrained foreign adversary whenever we build a regulatory fence around the open web or turn domestic cloud infrastructure into a gated defense site.
The true crisis of the artificial intelligence era is not that AI will become sentient; it is that in our rush to control the technology, we could dismantle our own constitutional privacy, surrender our software supply chains to un-auditable machine logic, and discover that we may be building a system that cannot be fully controlled nor switched off, creating the ultimate national security threat.
That said, we must realize that this is simply an awareness of trajectory; it is not a prophecy, and the difference between the two is everything. The same human ingenuity now racing to build these systems is fully capable of engineering the instruments to govern them. In our enthusiasm to create something that can help humanity in countless ways, we must also remain aware of the challenges of such advanced technology, at this speed of development.
ANTHROPIC ARMAGEDDON: DARIO AMODEI’S DOOMSDAY FEARS CONVENIENTLY CRUSH HIS COMPETITION
Artificial intelligence remains one of the most promising tools humanity has ever devised — already compressing years of medical discovery into months, advancing scientific exploration, broadening access to education once out of reach, and putting enterprise-grade capability within reach of the small businesses that could never before afford it, all of it widening the possibilities of what we can achieve.
We are right to be startled by AI’s capabilities, and yet that is exactly what makes it all so exciting: that step into a future of potential that those who came before us only dreamed of.
Jacqueline Cartier is a corporate and legislative strategist focused on communications, crisis leadership, public trust, and emerging technologies that shape human behavior and decision-making. Follow her on LinkedIn.