The U.S. Securities and Exchange Commission this week finalized a rule that requires public companies to disclose the material use of generative artificial intelligence in financial reporting, investor communications and other disclosures when such use could reasonably affect investors’ understanding of the company’s business, operations or financial condition.

What the rule requires

Under the new rule, companies must disclose when generative AI materially contributed to the preparation or presentation of:

  • Periodic financial statements, management discussion and analysis (MD&A) and other filings;
  • Earnings releases, investor presentations and analyst materials;
  • Automated trading or valuation models that rely on generated outputs used in pricing, forecasting or hedging.

Disclosures must describe the nature and extent of AI use, the types of models or services employed (vendor or in‑house), significant limitations or assumptions, and key controls applied to ensure accuracy and reliability. Companies also must maintain contemporaneous logs and documentation demonstrating how AI outputs were validated and incorporated into materials provided to investors.

Phased compliance and audit expectations

The SEC set a phased compliance timeline: an initial 12‑month implementation window for disclosure and basic documentation, followed by a 24‑ to 36‑month period during which registrants must establish and report on enhanced governance controls, model validation processes and third‑party attestations where appropriate.

Auditors will be expected to consider AI use in their inspections of financial reporting controls. The rule signals that external auditors may need to expand procedures to examine model documentation, testing records and access controls. The SEC’s release indicates enforcement will focus on material misstatements and failures to disclose AI reliance that mislead investors.

Why the SEC acted now

Regulators said the decision responds to the rapid adoption of generative AI across corporate functions — from automated drafting of earnings commentary to AI‑assisted forecasting and automated customer‑facing disclosures — and the potential for generated content to introduce errors, fabrications or materially alter investor expectations.

Market participants have increasingly used LLMs and multimodal generators to summarize data, draft narratives and produce forecasts. Regulators argued investors need clarity about when those outputs are material to a company’s public statements.

Immediate industry reaction

Large accounting firms and corporate legal teams have already announced advisory services geared to the new rule. Several audit firms said they will expand AI model review capabilities, including specialist teams to evaluate model governance, training-data provenance and validation results.

AI vendors responded by rolling out “investor‑grade” attestations for enterprise customers: standardized evidence packages that document model architecture, training and update history, performance metrics, safety testing and access logs. Some vendors also introduced fine‑grained export controls and tamper‑evident audit trails aimed at easing customer compliance.

Practical implications for companies

Chief financial officers and AI leaders should expect to do three things immediately:

  1. Inventory: Identify all material uses of generative AI across reporting, forecasting and investor communications.
  2. Control frameworks: Implement validation, testing and change‑management processes that are documented and repeatable.
  3. Disclosure playbook: Draft disclosure language tailored to different levels of materiality and prepare internal sign‑offs required before investor‑facing materials are published.

Legal and compliance teams will need to revise disclosure committees’ charters and update policies governing vendor risk management, data lineage and retention of AI output logs. Finance teams should integrate validation checkpoints into close and earnings cycles to ensure AI‑generated narrative or quantitative contributions are verified prior to release.

Sectors to watch

Regulated sectors where even small data errors can be material — financial services, pharmaceuticals, energy and public utilities — face the highest compliance burden. For example, banks using AI in credit‑loss forecasting or model risk management will need to document the role generative components play in forecasts and stress‑testing outputs.

Companies that depend on investor confidence in precise forecasts or audited metrics should expect close scrutiny and may opt for stronger attestations (including third‑party model validation) to reduce disclosure risk.

Open questions and next steps

The rule leaves several operational questions to industry guidance and SEC staff bulletins. Notably, the scope of “material” AI use will be tested in early enforcement cases; companies and auditors will look to the SEC for examples and clarifications on thresholds for disclosure.

Additionally, international firms listed in the U.S. will need to reconcile overlapping regulatory regimes where other jurisdictions have different AI reporting requirements. Cross‑border companies should map disclosure obligations and align documentation practices to satisfy multiple regulators.

Advice for AI teams

  • Start an immediate cross‑functional inventory of AI use cases tied to investor materials.
  • Establish model cards, test suites and lineage records for any system feeding financial statements or investor communications.
  • Work with legal to create templated disclosure language and with finance to add verification checkpoints to the earnings process.

The SEC’s new rule marks a turning point in how enterprises must govern generative AI when it intersects with investor information. For AI teams and corporate officers, execution will require tighter collaboration with finance, audit and legal functions — and a rapid buildout of documentation and control mechanisms that until now many organizations have not prioritized.