There's something darkly amusing about the current state of tech discourse. We're collectively wringing our hands over whether artificial intelligence will be "governed responsibly" while quietly allowing hardware manufacturers to build surveillance infrastructure into the most mundane devices in our homes. The incentive structure here is telling us exactly where corporate priorities lie, and we should be paying attention to who profits from our inattention.
Recent developments in WiFi identification technology, which can now recognize individuals with near-perfect accuracy through walls and clothing, represent a watershed moment we've mostly ignored. This isn't theoretical. This is deployable, now, in routers that millions of households already own. Yet the conversation around this capability has been muted compared to the reasonable anxiety we've directed toward large language models and their training practices.
Why the disparity? Follow the incentives.
AI companies face constant scrutiny from activists, policymakers, and media critics. The pressure is relentless, often warranted, and increasingly consequential for their bottom line. Tech firms have learned they must at least perform concern about responsible development. They hire ethics boards. They issue statements about alignment. Whether these efforts are genuine or performative, the point is that reputational risk now attaches to AI development in ways that make these companies cautious.
WiFi hardware companies operate in a different ecosystem. They're not household names. Most people couldn't name the chipset manufacturer in their router. The business model is straightforward: sell hardware, occasionally push firmware updates, collect whatever data the device generates, sell or monetize that data through various channels. Surveillance capability isn't a bug they're trying to fix. It's an asset they're trying to quietly operationalize.
Here's the uncomfortable part: the incentive structure actually rewards the WiFi approach. There's no reputational downside to quietly adding identification capabilities to a consumer device. The technical barrier to public awareness is high. The regulatory environment is still playing catch-up. Meanwhile, the potential value of knowing who's in a building, when they're there, and how they're moving through space is enormous. Insurance companies. Retail analytics. Real estate firms. The security industry. All would pay for this capability.
The contrast with AI governance isn't accidental. We've created a situation where the most visible, scrutinized technologies face enormous pressure to be "responsible," which actually pushes investment and capability development toward less visible technologies that operate in regulatory and cultural blind spots. A company building a language model faces public criticism and policy pressure. A company integrating identification into WiFi infrastructure faces... what, exactly? A few technical articles most people won't read.
This pattern repeats across the technology sector. We obsess over the thing everyone's discussing while the more consequential capability development happens in the margins. It's not necessarily because the marginal technologies are more dangerous. It's because the incentive structures point different directions.
The real problem isn't that we're paying too much attention to AI governance. The real problem is that we've created regulatory and reputational environments where visibility and scrutiny are treated as sufficient controls. They're not. They just redirect development toward less visible channels.
What should concern us: a future where the most powerful surveillance capabilities aren't controversial because they're not famous. Where body-identifying WiFi routers become standard issue because no one organized to prevent it. Where the tech that shapes our lives operates in the spaces we're not looking.
The industry is rewarding invisibility. We should notice who's building in the dark.