SEC asks how machines read the market
The US Securities and Exchange Commission (SEC) has spent a day exploring a question that has never been so relevant to XBRL filing: what do machines need from corporate disclosure? Its public roundtable on machine-understandable data in the age of “super intelligence” was organised by the Office of Data Standards and Innovation within the Division of Economic and Risk Analysis (DERA). It brought panellists from Meta, Google, BlackRock, JP Morgan Chase and Two Sigma together with data providers, auditors and academics.
Chairman Paul Atkins closed the session by noting that the SEC now offers 15 data sets free of charge, and that use of its Financial Statement Data Set has grown more than fivefold in two years. He also commented that “better data need not mean more disclosure,” arguing that the greater opportunity lies in making what companies already disclose easier to use, without adding to the burden on preparers. New tools do not, however, let issuers off the hook, as weak data limits what even the most sophisticated analysis can do.
The Financial Statement Data Set is extracted from the XBRL reports companies file with the Commission, so a fivefold jump in its use is, in effect, a fivefold jump in the use of XBRL data. This lines up with the download figures Chief Economist Joshua White shared in September, when he told an industry audience that machines are now one of the two audiences for financial data.
Those machines need high-quality, well-structured, and trustworthy data if they are to reach their full potential. “Consider, for example, a high-performance engine. However powerful or finely engineered, it is only as good as the fuel that goes into it,” said Atkins.
The Chairman’s comments set out a clear position: standardised, machine-readable data can make disclosure more useful without making it more arduous. XBRL International is in full agreement, and it is great to hear it from the Chairman’s lectern.
Read Chairman Atkins’ remarks here.

