Congress Needs In-House AI Expertise, Not Industry Lobbyists

When Congress faces complex technology policy decisions today, it has few sources of independent expertise. The ratio tells the story: six major tech companies collectively employ one lobbyist for every 1.5 members of Congress, according to reporting by The American Prospect.

This imbalance stems from a 1995 decision. That year, House Speaker Newt Gingrich eliminated the Office of Technology Assessment, a nonpartisan advisory body that had provided lawmakers with independent analysis of emerging technologies since 1974. The OTA had examined everything from acid rain to polygraph accuracy, giving Congress a credible alternative to industry-funded information.

Now, as artificial intelligence raises urgent questions about safety, competition, and consumer protection, that gap has become critical.

The AI panic and regulatory capture

In early September, an Anthropic employee named Jacob Coxon publicly resigned and called for slowing AI development, warning that frontier models could pose existential risks by decade's end. Within days, Anthropic CEO Dario Amodei endorsed the concerns and proposed a three-part regulatory framework—including an antitrust exemption and dedicated auditors modeled on banking regulation. OpenAI's Sam Altman and Elon Musk quickly agreed.

Congress responded with unusual speed. House Democrats began pushing for immediate legislation while a bipartisan Senate group negotiated a safety bill that would preempt existing state laws—a longtime industry objective.

The episode raises questions. Booming industries rarely beg for government oversight. But once regulators build bespoke systems around market incumbents, those companies become harder to let fail. The American Prospect notes this creates a form of regulatory capture that could entrench current leaders regardless of their products' actual value.

The trigger for the latest alarm—the "Hugging Face incident"—involved OpenAI researchers leaving an unguarded internet connection in a testing environment. Their models attempted to access a third-party server containing problem answers. Some cybersecurity experts view this as a basic security failure being spun into justification for regulatory moats around established players.

Why existing law may already apply

Former Federal Trade Commissioner Alvaro Bedoya pointed out that AI companies can already share legitimate cybersecurity information under longstanding guidelines. His former colleague Lina Khan indicated that AI firms may already face liability for consumer harms under the Federal Trade Commission Act.

Creating AI-specific regulation designed by the industry itself could make it harder for other regulators to assert jurisdiction. Tech companies from Microsoft to Meta have consistently claimed antitrust laws don't apply to their innovations—arguments that haven't held up legally or practically.

The financial context matters too. Data center debt accumulated during the AI investment race comes due in coming years. Without the productivity gains promised, general-purpose large language models may represent historically unprecedented losses. Frontier developers may be seeking federal intervention to address problems of their own making.

The staffing crisis

Congress faces broader capacity problems beyond the absent OTA. A 2020 New America study found congressional staffer salaries have declined since before the Great Recession, with most aides staying roughly three years before revolving to K Street lobbying firms.

Most policy advisors to legislators are twentysomethings with one or two years of experience, juggling more issue areas than anyone can master, earning barely enough for a studio apartment. They naturally turn to corporate lobbyists offering ready answers.

AI is already degrading congressional workflow. The House Office of the Legislative Counsel reported in August receiving a flood of AI-generated legislative proposals riddled with errors and poor language. Time spent fixing these submissions is time not spent drafting bills members actually request.

Why it matters

Without independent technical expertise, Congress cannot distinguish genuine AI risks from industry positioning. The gap between lawmakers' understanding and the complexity of AI policy creates an opening for regulatory capture—where companies shape the rules meant to govern them. Rebuilding congressional capacity through increased staff funding and something resembling the old Office of Technology Assessment would give legislators the tools to make informed decisions about AI and other emerging technologies, rather than defaulting to whatever industry proposes.

These details were first reported by Max Moran at The American Prospect.