Companies are Using AI More Than Ever. Can Their Formal Reporting Keep Pace?

AI is becoming a strategic necessity for many companies in the economy, including both developers and deployers of AI systems. AI has the potential to transform business models and operations, and reshape industries such as banking, healthcare, and retail.

As corporate AI use spreads, decision-makers need insight into how the potential impacts of AI — such as bias, privacy, and emissions — interact with companies’ ability to create value for investors and other stakeholders. Adopting and commercializing AI responsibly requires risk-informed innovation that considers how AI’s impact on people and the environment affects business opportunities.

For this reason, investors, regulators, and analysts should receive high-quality information about the impacts, risks, and opportunities of AI in formal financial and sustainability reports, just as they do now on topics such as cybersecurity, supply chain resilience, and human capital.

Today, Partnership on AI is publishing Disclosure of AI-related impacts, risks, and opportunities, a landscape analysis of the current state of formal reporting. We reviewed 50 of the world’s largest companies from a variety of industries and assessed the quantity and quality of AI disclosures in their formal reports. The landscape analysis highlights the need for more consistent, comparable, and decision-useful information.

Formal reports aren’t marketing materials, but resources to help decision-makers assess a company’s financial health, societal impacts, and strategy over the short, medium, and long term.

Formal Reporting as Essential for Informed Decision Making

A foundational concept in formal reporting is that investors and other stakeholders require high-quality information for informed decision-making. Investors, policymakers, and the public rely on these reports to understand how a company creates value and manages risk.

Formal reporting differs from other communications because it follows certain principles—such as consistency, comparability, and faithful representation—and complies with well-established disclosure standards. These aren’t marketing materials, but resources to help decision-makers assess a company’s financial health, societal impacts, and strategy over the short, medium, and long term.

An effective formal report should disclose information that is material to readers and could influence the decisions of investors, regulators, or the public.

For primary users of financial reports, such as investors, lenders, and other creditors, this includes financial performance, cash flow, and the cost of capital. For primary users of sustainability reports, such as civil society organizations, governments, and academics, this includes impacts on people and the environment.

In the case of AI, a core set of impacts on people and the environment that are material to society—such as privacy, bias, and safety—may also be material to investors. These two aspects can be hard to separate, and information that is important to other stakeholders may also interest investors.

The Current Landscape

In our landscape analysis we examined financial and sustainability disclosures from 50 major companies—25 technology companies (e.g., AI developers, enterprise software providers, and chip manufacturers) and 25 companies from other industries (e.g., financial services, healthcare, automotive, retail, and entertainment) to determine whether the material impacts, risks, and opportunities related to AI systems are sufficiently disclosed today.

We analyzed formal financial reports (e.g., SEC Filings), formal sustainability reports, and formal AI-specific reports. We did not review informal communications such as blogs, websites, and public relations materials.

We found that the quality, quantity, and location of disclosures vary significantly, and while best practices have emerged, significant gaps remain in the disclosure of AI-related impacts, risks, and opportunities. This finding reflects the reality that mainstreaming AI is a relatively recent phenomenon and has yet to benefit from the continuous improvement process that has enhanced other elements of sustainability and financial reporting over many decades.

Sustainability reporting standards in use today, such as those from the International Sustainability Standards Board (ISSB), the Global Reporting Initiative (GRI), and the European Sustainability Reporting Standards (ESRS), have improved through extensive testing, stakeholder input, and refinement. As a result, they offer clear, consistent, and compatible frameworks to improve the quality of reporting on AI-related impacts, risks, and opportunities.

AI innovation and adoption relies on investor and stakeholder trust.

Emerging Best Practices and Gaps

In reviewing the landscape analysis, several clear priorities emerged:

• Consistency and comparability: AI disclosures today cover many similar topics (e.g., bias, privacy, liability, security), but the substance of disclosures varies significantly from company to company. New guidance can improve the consistency and comparability of disclosures.
• Location of disclosure: Companies often make high-level disclosures about business risks in their SEC filings and high-level disclosures about policy, governance, and impacts in their sustainability reports; some companies provide more detail in dedicated AI reports. Still, the location of disclosure varied significantly within and between reports, and the analysis of AI impacts, risks, and opportunities could be improved with more harmonized disclosure locations and increased connectivity between them.
• Best practices for disclosure quality: AI disclosures today cover a much broader range of topics than ever before, but additional insights—such as trends, likelihood, and relevance for business performance—can make disclosures more decision-useful.
• Quantitative metrics: Almost all AI disclosures regarding impacts, risks, and opportunities of AI for people and society are presented as qualitative information, such as principles, commitments, policies, processes, and programs. Effective reporting requires a mix of numbers and narrative, so the further development of corporate-level quantitative metrics will be helpful.
• Linking AI systems to corporate-level risk: Significant progress has been made over recent years in developing risk assessment frameworks and benchmarks for AI models and systems, such as model or system cards. It would be helpful to consider whether mainstream financial and sustainability reports should include information about the results of AI model and system risk assessments and benchmarking, or whether this information should be kept separate from corporate-level reporting.

We also found several topics with less disclosure than might be expected given what we know about AI, such as its potential impact on the workforce, the environmental impacts of developing AI, and the risks and opportunities arising from downstream AI use. While each company should determine its own material topics, we expect that report users are seeking enhanced disclosure from companies on these issues, which may also vary by industry.

We held discussions with both responsible AI and reporting experts to inform the landscape analysis. We discovered that responsible AI experts are eager to understand how well-established formal reporting standards can capture the impacts, risks, and opportunities of AI, and reporting experts want to explore how reporting standards should evolve to reflect AI’s transformational significance.

Both communities recognize the strategic importance of AI for companies and the significance of its impacts on people and the environment. As companies, regulators, and standards-setting bodies seek to enhance the quality of formal reporting in AI impacts, risks, and opportunities, it will be essential to bring these different professional communities together. At PAI, we are committed to advancing this work with our partners around the globe, convening practitioners and experts in both fields to shape practical AI disclosure guidance.

Ultimately, AI innovation and adoption relies on investor and stakeholder trust. Companies that proactively assess and disclose the impacts, risks, and opportunities of AI are more likely to earn the confidence of their customers, partners, and regulators, inform improved decision-making by their investors, and support the improved alignment of capital and strategy.

If you are interested in contributing to our work or learning more about our findings, please contact us at contact@partnershiponai.org.