UpTrajectory Review

Lina Khan, the former FTC chair who became the face of aggressive antitrust enforcement under Biden, went on ABC's This Week to deliver a blunt warning: letting AI companies regulate themselves would repeat what she called the 'proven failure' of Big Tech's self-regulation during social media's rise. Her argument is historical, not speculative. A decade ago, platforms promised to handle misinformation, privacy, and harm voluntarily; Khan's view is that the record since then, from Cambridge Analytica to teen mental health concerns, demonstrates what self-regulation actually produces when shareholder incentives and public safety collide.

For a small-business operator, this is not an abstract policy debate. The AI tools you're adopting, whether for customer service, marketing copy, hiring screens, or inventory forecasting, are built by companies operating in what Khan describes as a regulatory vacuum. If self-regulation fails the way she predicts, the compliance burden won't fall on Microsoft or Meta first; it will cascade downstream to every business that embedded their tools without contractual protections, audit trails, or contingency plans. Operators who treat AI vendors like any other software supplier, without interrogating data practices or model behavior, are accepting risks they haven't priced.

What makes Khan's intervention notable is the timing and the unusual alignment it reveals. Anthropic CEO Dario Amodei published a blog post last month explicitly calling for federal regulation targeting frontier AI companies, a position that drew public support from both Sam Altman and Elon Musk, competitors who agree on almost nothing else. When the CEOs of the three most prominent AI labs are asking to be regulated, Khan's skepticism of voluntary agreements carries more weight. The Trump administration, however, has signaled the opposite direction: Trump announced last week that tech leaders signed a self-regulation pact, which Khan's framing directly undercuts.

The second-order effects cut in multiple directions. If Khan is right and harms accumulate, small businesses face a familiar pattern: regulation arrives late, retroactively, and with compliance costs that scale with company size, meaning large incumbents absorb them easily while smaller operators struggle. If she's wrong and self-regulation works, businesses that over-invested in compliance infrastructure wasted capital they could have deployed elsewhere. Either way, the uncertainty itself has a cost. Businesses making multi-year AI investments right now are effectively betting on which scenario unfolds, and most are making that bet without realizing it.

Watch two things over the next six months. First, whether any state, California and Colorado especially, moves to fill the federal gap with AI legislation that imposes actual obligations on vendors, because state-level fragmentation would create a compliance patchwork that hits multi-state small businesses hardest. Second, whether the voluntary safety agreement Trump announced includes any enforcement mechanism or reporting requirements, or whether it's a press release. In the meantime, operators should be asking AI vendors harder questions: what data trains your models, what happens to our inputs, and what liability do you accept if the tool produces something harmful.

“We've seen that self-regulation efforts by Big Tech have been a proven failure.” — Business Insider

Takeaway: Treat AI vendors like unregulated utilities: demand contractual data and liability terms now, before a self-regulation failure forces reactive compliance costs onto your business.