# Opinion: Silicon Valley owes America more than its weapons > Source: > Published: 2026-09-15 10:30:45+00:00 **Getting your** [Trinity Audio](//trinityaudio.ai)player ready... Last month, Meta agreed to pay up to $17.1 billion to 47 states, D.C., and territories over claims it designed its platforms to addict children and misled the public about its own research, the largest such settlement outside tobacco. Meta admitted no wrongdoing and agreed to pay the full amount only if its competitors adopt the same teen safety measures. Three weeks earlier, a New Mexico judge ordered Meta to pay $942 million for what the court called a “public nuisance.” Days later, Zuckerberg published a 6,500-word manifesto calling for broad access to superintelligence and fewer constraints on its development. It mentioned neither the verdict nor safety. That gap is not an oversight, it is posture. Silicon Valley’s leaders are articulate about what technology will do for humanity — and much less consistent about what the industry owes the society that made it possible. Alex Karp argues in The Technological Republic that the sector’s debt to America is paid through defense. Dario Amodei writes of AI’s civilizational promise. Zuckerberg promises widely distributed superintelligence. Each contains something true. Together they leave out much of the bill. I spent years at that intersection: in uniform, at the White House, in agencies working where national security and emerging technology collide, watching the struggle to keep pace with capabilities the industry deployed faster than any governance structure could absorb. Warfighters deserve every advantage technology can provide, but this does not settle the account. Silicon Valley did not build itself. Its foundations lie in taxpayer-funded DARPA grants, public universities and decades of national-security investment. That history creates an obligation, and this 250th year of the republic is the right moment to ask what it includes. We’ve witnessed the impact of capabilities arriving far faster than governance. The FBI recorded more than 22,000 AI-related fraud complaints last year, nearly $900 million in losses: voice clones draining retirees’ savings, and billions laundered through cryptocurrency rails. Software shipped at scale — insufficiently tested, insecure by default, liability disclaimed — has become the battlefield adversaries exploit against hospitals, water systems and municipalities. The costs of technological scale fall, as they always do, on people least able to absorb them. The technology is now moving into territory where assumptions about human control are harder to sustain. Recent incidents have shown AI models reaching external systems and taking actions beyond the boundaries researchers intended. In one case, a model escaped a controlled testing environment and carried out thousands of actions against an outside system; subsequent reviews found similar containment failures. The rules for testing, containing and deploying systems with this kind of autonomy cannot be written by companies alone. Rather than develop a sensible framework, federal policy has moved in the opposite direction. The administration dismantled the AI safety framework established by the previous administration. Its June executive order on frontier models relies on voluntary engagement rather than mandatory requirements. This month it authorized private companies to conduct offensive cyber operations against criminal networks — without the accountability frameworks that determine responsibility when operations go wrong, including implications for treaty obligations, due process and the laws of armed conflict. Silicon Valley’s most powerful technologies are increasingly invited to serve as instruments of national power. The obligations to manage their consequences remain elective. This is not novel. High-risk technologies undergo independent evaluation before going to market; AI and other high-impact emergent tech should be no different. We need a standing, independent capability to investigate serious AI incidents, closer to the National Transportation Safety Board than a traditional agency, so failures produce public lessons rather than corporate postmortems. A FINRA-like oversight body could provide continuing supervision, provided its independence, funding and model access are guaranteed by law. Accountability should follow choices that foreseeably cause public harm; disclaiming liability in fine print or passing the buck to end users will not do. AI platforms and hyperscale data-center operators should fund worker transitions and grid resilience up front, like infrastructure performance bonds, before those costs land on the public. None of this suppresses innovation; it builds the institutions capable of sustaining it. I served this country because I believed the republic was worth defending — not as an abstraction, but as the actual lives of actual people: their savings, their children’s safety, their ability to trust what they see and hear and read. The companies now claiming the mantle of tech patriotism were built on the same public investment that trained the people who served alongside me. Paying that debt means accepting that the power to reshape how Americans live, work and govern themselves carries an obligation to sustain what makes that life worth living. The founders built courts, universities and a free press for the same reason. The question now is whether American technology will do the same. That is the tech patriotism worth believing in. *Carole House is a former White House National Security Council special advisor, co-author of President Biden’s executive orders on digital assets and cyber innovation, former chair of the CFTC Technology Advisory Committee, member of Governor Newsom’s Innovation Council and senior fellow at the Atlantic Council GeoEconomics Center.*