---
title: "Does The AI Industry Need Its Own FINRA?"
url: https://stacklist.com/card/511e0de3-55fb-44f1-a576-ff16c5f7def8
source_url: "https://www.forbes.com/sites/jamesbroughel/2026/07/19/does-the-ai-industry-need-its-own-finra/"
stack: https://stacklist.com/c/finance/stack/f947deb3-cc46-4708-be70-b5d70fe6a5d3
summary: "The article examines the Trump administration's proposal to create a FINRA-like independent body to vet advanced AI models for safety before public release, highlighting both the appeal and the criticisms of using FINRA as a regulatory template. Critics argue that FINRA's accountability gaps and its poor track record during the 2008 financial crisis suggest policymakers should leverage existing voluntary AI standards frameworks instead of building a new quasi-governmental bureaucracy."
tags: "ai-regulation, finra, ai-safety, trump-administration, oversight, artificial-intelligence, policy-analysis"
key_entities: "FINRA (organization), James Broughel (person), Scott Bessent (person), Trump Administration (organization), Anthropic (organization), Demis Hassabis (person), Google DeepMind (organization), Securities and Exchange Commission (organization), Fable (technology), Mythos (technology), AI oversight (concept), self-regulatory organization (concept), Forbes (organization), Commerce Department (organization)"
classification: "analysis"
content_hash: "sha256:5d915796096f6501d3e36c0d122edca86af7160ae88399effe6b0b7f54cb4378"
acp_version: "0.2"
token_counts_approximate: 4134
visibility: public
agent_accessible: true
status: "final"
---

# Does The AI Industry Need Its Own FINRA?

Today’s Stocks Money Markets Does The AI Industry Need Its Own FINRA? By James Broughel , Contributor. Forbes contributors publish independent expert analyses and insights. James Broughel is an economist focused on the economics of regulation. Follow Author Jul 19, 2026, 05:57am EDT Jul 19, 2026, 06:43am EDT --:-- / --:-- This voice experience is generated by AI. Learn more . This voice experience is generated by AI. Learn more . Summary The Trump administration is reportedly considering a FINRA-like independent body to vet advanced AI models for safety before public release. This proposal aims to standardize oversight, moving beyond ad-hoc interventions that have frustrated AI companies. Proponents highlight FINRA's ability to attract talent and adapt quickly. However, critics cite FINRA's significant accountability gaps and its problematic record during the 2008 financial crisis. Creating a coercive new regulator overlooks the robust, voluntary AI standards ecosystem already in place. Policymakers are urged to consider existing frameworks, rather than creating a new, potentially flawed, quasi-governmental bureaucracy. Show More The Trump administration is weighing a new AI oversight body modeled on FINRA, the private regulator that polices the brokerage industry. getty Bloomberg reported last week that the Trump administration is weighing the creation of an independent body to vet the safety of advanced artificial intelligence models before they reach the public. The proposal would establish a new oversight organization modeled on the Financial Industry Regulatory Authority, known as FINRA, the private regulator that polices the American brokerage industry under the supervision of the Securities and Exchange Commission. Frontier AI labs would submit their most capable models for a thirty day review before release, with experts screening for cyber, biological, and deception risks. Participation would begin on a voluntary basis and could later become mandatory once testing protocols prove themselves. Treasury Secretary Scott Bessent, who has taken an active role in the administration's AI policy this year, reportedly helped develop the concept, which remains under review at the White House. The proposal lands after weeks of confusion surrounding Anthropic’s Fable model and its more powerful sibling Mythos. Both were already on the market in June when a Commerce Department export control order froze them overnight amid national security concerns. Anthropic then spent weeks negotiating the models’ return with no established rules or procedures to guide the process, and both eventually came back online. AI companies have complained about the ad hoc character of these recent federal interventions in model releases, and Google DeepMind chief executive Demis Hassabis has publicly called for an American led oversight body along the lines of the Bessent plan. Yet before Washington borrows FINRA as a template, policymakers should examine how the original agency has actually performed. FINRA’s institutional design has drawn sustained criticism from serious observers, including from within the SEC itself. That record suggests the FINRA model may not deliver what its admirers expect. Why The FINRA Model Appeals To Policymakers The attraction to FINRA is understandable. Frontier AI models are among the most complex software systems ever built, and evaluating them requires talent that traditional agencies struggle to recruit and retain. A body funded by industry fees, rather than congressional appropriations, could pay competitive salaries and staff itself with engineers and researchers who understand the systems they are reviewing. FINRA itself oversees roughly 3,250 securities firms and more than 630,000 registered representatives, with a workforce of specialists the SEC could never house on its own budget. Speed is a second advantage. A private body can update its rulebook faster than an agency bound by the full administrative process, which matters in a field where model capabilities change every few months. A third is predictability. Labs would know who reviews their models, against what criteria, and on what timeline, replacing a regime in which release decisions can depend on improvised interventions from the White House. Industry participation gives companies a voice in shaping the standards they must meet, which encourages cooperation. These are real advantages. MORE FOR YOU FINRA's Accountability Gap Few people have scrutinized FINRA more carefully than SEC Commissioner Hester Peirce. In a 2015 study for the Mercatus Center, written before she joined the Commission, Peirce argued that FINRA no longer provides self regulation in any meaningful sense. Its board is deliberately weighted against the industry, with a majority of governors required to have no industry ties, so member firms exercise limited control over the organization that regulates them. This can help prevent regulatory capture, but it also weakens some of the traditional advantages of industry self-regulation. At the same time, FINRA escapes the mechanisms that hold government regulators accountable. It faces no notice and comment obligations under the Administrative Procedure Act, no Freedom of Information Act requests, no congressional appropriations oversight, and no requirement to weigh the costs and benefits of its rules. Peirce concluded that FINRA “wields governmental powers without the procedural and disclosure requirements by which a government regulatory agency would be constrained.” SEC oversight exists on paper. The Commission approves FINRA rules and can review its disciplinary decisions. In practice, Peirce found, FINRA sets its own agenda and budget without SEC input and operates with substantial independence. Nor does competition discipline the organization. FINRA itself was born of consolidation. In 2007 the National Association of Securities Dealers, which had supervised brokers since 1939, merged with the regulatory arm of the New York Stock Exchange to form FINRA. That merger left a single self regulator standing, and FINRA has been the only game in town ever since. Notably, Peirce has long been sympathetic to self regulation as a concept. Her complaint is that FINRA occupies an uncomfortable middle ground, neither a true self regulator run by the industry nor a government agency answerable to voters. Among the reforms she proposed was allowing competing self regulatory organizations to emerge. What The Financial Crisis Revealed Institutional design flaws might be forgivable if the regulatory model produced good results. On this front, FINRA’s record should give us pause. The organization opened its doors in July 2007, mere months before the financial system began to unravel. But it was hardly a novice. It inherited decades of supervisory experience from the NASD and the NYSE regulatory unit it absorbed, along with responsibility for the industry those predecessors had been watching. The firms under that supervision included Bear Stearns, Lehman Brothers, Merrill Lynch, Madoff Investment Securities, and Stanford Financial Group, a roster that reads like a laundry list of the era's failures. The Madoff case drew particular scorn. FINRA maintained that the fraud occurred in Madoff's investment advisory business, outside its jurisdiction. But securities law scholar John Coffee testified before the Senate Banking Committee that Madoff's brokerage operation fell squarely within FINRA's jurisdiction, and that the brokerage firm FINRA examined for years was the same firm that supposedly held custody of the assets Madoff's victims thought they owned, assets that did not exist. A scandal surrounding auction rate securities was harder still to explain away. Auction rate securities are long term bonds whose interest rates reset at auctions held every few weeks. Wall Street firms marketed them to customers as safe, highly liquid investments that were as good as cash. That pitch held up only as long as the auctions functioned. When brokerages quietly stopped bidding in the auctions, as they had done for years, the entire market froze within days, and tens of thousands of investors discovered they could not sell what they had been told were cash equivalents. FINRA itself got out in time, however. As a Project On Government Oversight letter to Congress recounted, the regulator sold its own $647 million portfolio of auction rate securities in 2007, just months before the freeze in February 2008. Meanwhile, investors who received no warning were left with tens of billions of dollars locked in a frozen market. FINRA said the sale reflected routine portfolio management, and no wrongdoing was ever established. Yet every reading of the episode is unflattering. Either the regulator’s investment arm benefited from what its examiners could see coming, or its own money managers judged too risky the very instruments its regulatory program allowed brokers to keep selling as safe. And because FINRA answers no Freedom of Information Act requests and courts have shielded self regulatory organizations from suits over their regulatory conduct, outsiders have no way to establish which it was. Fairness requires acknowledging that blame for the crisis was widely shared beyond FINRA. The SEC missed Madoff’s Ponzi scheme too, despite repeated and detailed tips. Banking regulators missed the mortgage risks accumulating on the balance sheets they examined constantly. FINRA existed to enforce rules about sales practices and broker conduct, not to monitor firm solvency. Still, critics argue the organization excelled at checking compliance boxes while missing the systemic problems building inside its largest members. FINRA proves an organization with broad authority, a large budget, and an army of examiners is no guarantee of effective oversight or catching catastrophic risks before they materialize. A Regulator Is Not The Same As A Standards Body There is also a deeper conceptual problem with the emerging debate. Coverage of the proposal has described the new AI agency interchangeably as a standards body and as a regulator. Hassabis speaks of an international standards organization, while the Bloomberg reporting describes an entity with the power to block model releases. These are fundamentally different institutions, and conflating them muddies the choice policymakers face. Industry standards organizations rest on voluntary participation. Their standards bind no one unless a government or customer separately chooses to require them. They pool technical expertise, develop standards by consensus, publish benchmarks and best practices, promote interoperability, and sometimes certify compliance. Firms adopt their standards because customers, insurers, and business partners demand it, a market driven mechanism that rewards standards for being useful. Self regulatory organizations in the FINRA mold are different creatures. Membership is a legal precondition for doing business. Their rules carry the force of regulation. They inspect member firms, bring enforcement actions, levy fines, and can expel a firm from the industry entirely, all under the umbrella of a government agency. The figure below summarizes the contrast. Industry standards organizations and FINRA style self regulatory organizations perform very different functions. Figure by author. Coercive power changes incentives. A standards body cannot keep a product off the market. A FINRA for AI eventually could. That power has an obvious use when a model is genuinely dangerous. But this permission slip model of regulation has serious downsides as well. A single gatekeeper concentrates risk. If its testing regime develops blind spots, there is no competing reviewer to catch what it misses. And the history of drug regulation shows that regulators are systematically risk averse. A reviewer who approves a model that later causes harm faces headlines, hearings, and blame. One who delays a beneficial model faces nothing, because benefits that never arrive are invisible. The Standards Ecosystem AI Already Has What this debate also overlooks is that AI already has a dense and active ecosystem of standards organizations, some governmental and some private, all of them voluntary. The National Institute of Standards and Technology released its AI Risk Management Framework in January 2023, a voluntary framework that companies and government agencies around the world now use to structure how they identify and mitigate AI risks. The International Organization for Standardization and the International Electrotechnical Commission published ISO/IEC 42001 in December 2023, the first international standard for AI management systems, and firms can already obtain third party certification showing they meet it. The joint ISO and IEC subcommittee on artificial intelligence has published dozens of AI standards covering everything from bias to risk management. IEEE's 7000 series of standards began with a process for addressing ethical concerns during system design and now extends to transparency, algorithmic bias, and data privacy. Newer and more specialized groups complement these older bodies. MLCommons, an industry benchmarking consortium, developed its AILuminate benchmark to grade model safety across twelve hazard categories, with participation from leading labs, academics, and civil society groups. The Frontier Model Forum, founded in 2023 by Anthropic, Google, Microsoft, and OpenAI, develops safety evaluations and shares research on the most capable systems. The Partnership on AI , a nonprofit coalition of academic, civil society, industry, and media organizations, convenes its partners to develop responsible practice guidance. Together these bodies already perform most of the functions the administration’s proposal contemplates. They develop evaluation methodologies, publish technical standards, encourage interoperability, disseminate best practices, and give researchers and competitors a venue for collaboration. Standard setting is especially valuable in a field like AI because the relevant knowledge lives inside the labs rather than in Washington. Voluntary consensus processes can iterate quickly, and mistakes get corrected by revising a document rather than by amending a binding rule. When standards compete, better approaches win adoption and weaker ones fade. Government can then build on that foundation, for example by writing procurement requirements or disclosure rules that reference established standards, without creating a new enforcement bureaucracy. Try The Market Before Mandates In a field as technically complex and fast moving as artificial intelligence, standards organizations are an obvious institutional response. They encourage learning, experimentation, and coordination while remaining flexible enough to keep pace with the technology. It is far less obvious that AI requires a FINRA style regulator endowed with formal authority and backed by the SEC, an agency whose expertise relates to securities markets rather than machine learning. Skeptics will point to an obvious gap in the voluntary approach. A standards body cannot stop a truly dangerous model from shipping. But the government is not powerless here, as the Fable episode demonstrated. Export controls froze the most capable models in the country overnight, and authorities of that kind remain available whatever AI oversight regime emerges. Those powers are blunt, and their use in June was clumsy. The remedy is clearer procedures around the authority that already exists, not a new institution with a gate through which every model must pass. The administration deserves credit for seeking a coherent alternative to ad hoc oversight, and its instinct to involve industry expertise is sound. But FINRA’s own history counsels caution. Before creating another quasi governmental regulator, policymakers should ask whether the standards ecosystem AI already has can accomplish most of the same objectives with more accountability and less bureaucracy. The proposal is worthy of the debate it is starting to receive. That alone does not make it the best institutional choice. Disclosure: The author was a colleague of Hester Peirce at the Mercatus Center and edited a book with her on the Dodd-Frank financial reform law before her appointment to the SEC. Editorial Standards Reprints & Permissions LOADING VIDEO PLAYER... FORBES’ FEATURED Video