SEC Holds Roundtable on AI-Readable Financial Data and Market Risks

The Securities and Exchange Commission is holding a roundtable in Washington on Monday to examine how financial-market participants use machine-understandable information alongside advanced artificial intelligence, and what risks and costs may follow. The November 9 gathering brings together Wall Street firms, technology specialists, accounting professionals and investor advocates, with a particular focus on the quality of data that automated systems depend on.

The SEC calls the event “Machine-Understandable Data in the Age of Super Intelligence (SI).” That description reflects the agency’s framing of the discussion, not a finding that any particular technology has achieved superintelligence. The central questions are more practical: how investors and businesses use structured financial information today, how reliable the resulting analysis can be, and whether the regulator could do more to make the information useful.

According to the SEC’s published roundtable agenda, the public program runs from 9 a.m. to 3:30 p.m. Eastern time at its headquarters at 100 F Street NE. An initial presentation on data sets is followed by two panels, one on market uses of machine-understandable data and the other on risks, costs and possible SEC support. The agency is also making a webcast available to viewers who do not attend in person.

Joshua White, director and chief economist of the SEC’s Division of Economic and Risk Analysis, said when the event was announced on October 19 that feedback from market participants could help staff understand how the investing public uses the technology. The SEC has not presented the roundtable as a vote on a new rule, and its published materials do not announce a specific regulatory change.

How financial firms use data that computers can read

The morning program begins with opening remarks involving Commissioner Mark Uyeda, whose contribution is listed as a recording, and White. Bob Luby of the SEC’s Office of Data Standards and Innovation is then scheduled to discuss data sets and their use before the first panel starts at 9:45 a.m. The sequencing puts the underlying information ahead of the applications built on it.

That first panel includes representatives of BlackRock, JPMorgan Chase, S&P, PwC and Two Sigma, alongside Hudson Labs and Suffolk University. Julie Marlowe, an assistant director in the SEC’s Office of Data Standards and Innovation, is the moderator. The participants span investment management, banking, market information, accounting and software-driven analysis, all areas in which the way financial information is organized can affect how easily it is used.

Machine-readable reporting is not new at the SEC. The regulator already requires various types of filings to carry structured data, often using Inline XBRL, a format that allows the same document to be read by people and processed by software. A tagged figure can identify not just an amount, but what the number represents, its reporting period and its accounting context. That is a different task from asking an AI model to infer figures from a page of ordinary text.

The distinction helps explain the roundtable’s emphasis. A system comparing revenue, cash balances or other reported measures across companies benefits when those figures are consistently labeled and can be traced to the original filing. The SEC says structured reporting can support large-scale analysis and comparisons across issuers and reporting periods. Such data can make an automated workflow more efficient, although a machine-readable label does not by itself establish that an underlying figure is accurate or comparable in every circumstance.

The agency also publishes data sets built from submitted filings and maintains tools for accessing information through its EDGAR system. Its economic-analysis division works on taxonomies, validation rules and data-quality checks. Those existing programs give the meeting a concrete policy setting: the SEC is asking how newer analytical technology interacts with a disclosure infrastructure that is already designed for computers as well as human readers.

Data quality, model reliability and the cost of adoption

The afternoon panel, scheduled for 1 p.m. to 3:15 p.m., turns explicitly to the risks and costs of using advanced systems with machine-understandable information. Valerie Szczepanik, listed on the SEC agenda as its chief super intelligence officer, is moderating. The discussion also asks how the Commission could better facilitate the use of these technologies, placing potential benefits and possible safeguards on the same agenda.

Speakers include representatives from Meta, Google, Workiva and financial-data provider Calcbench. The scheduled participants also include American University law professor Hilary Allen, Better Markets representative Evan LeFlore and assurance specialist Christina Ho. Their presence offers an opportunity for differing views on where automation improves access to information and where it may introduce problems. The agenda identifies these questions for discussion; it does not record the panelists as having agreed on particular answers.

A central concern is the gap between readable data and dependable analysis. Consistent formatting can help a system locate a disclosed number, but analysts still need to know whether it reflects the right company, accounting period and definition. A model that retrieves an incorrect figure or overlooks a qualifying note can produce a polished explanation that misstates the filing. These are potential failure points inherent in automated financial analysis, not findings the SEC has attributed to the participants in this roundtable.

Costs matter as well. Preparing disclosures for software use, checking the quality of tagged information and building tools to analyze it require resources. Large institutions may be able to maintain specialized data and engineering teams, while smaller firms and individual investors may rely on services supplied by others. The policy question is whether wider access to well-structured public information can narrow that gap, and what safeguards are appropriate when users cannot readily inspect how an automated answer was generated.

Inline XBRL illustrates both the opportunity and the limits. The format can display an ordinary filing while also embedding tags that software can interpret, and its EDGAR viewer lets readers inspect the context of individual data points. That traceability is valuable when financial information is reused in an AI-powered search, screening or research tool. It does not eliminate the need to review the source disclosure, especially when the decision involves risk, valuation or an investment recommendation.

What the roundtable may mean for SEC oversight

The event is a public discussion rather than a rulemaking proceeding. The agency’s announcement set out questions for market participants, but did not release proposed regulatory language, a compliance deadline or a Commission vote. Any future initiative would need to be evaluated on its own terms; the existence of a roundtable should not be read as evidence that new AI disclosure or reporting requirements are imminent.

Attendees also approach the subject from different vantage points. Financial firms may focus on extracting comparable information from reports, software providers on data access and usability, and investor advocates on the reliability and availability of analysis. Those are different interests, and improvements for one group may not address the concerns of another. Gathering those perspectives can help SEC staff distinguish practical obstacles from problems that might warrant a policy response.

Structured reporting has a regulatory history at the SEC. It adopted amendments in 2018 requiring Inline XBRL for specified corporate financial information and fund disclosures on a phased basis, and the technology now supports a range of filing requirements. Monday’s roundtable does not reopen those rules automatically. It instead examines how the use of the data has evolved as increasingly capable software processes public-market information at scale.

Chairman Paul Atkins is scheduled to deliver closing remarks from 3:15 p.m. to 3:30 p.m. The SEC says it will make a recording available after the live webcast. Until the proceedings and any subsequent agency statements are available, the verified development is the roundtable itself and the issues on its published agenda, not a settled regulatory outcome.