Law and the Emerging Political Economy of Algorithmic Audits

Overview

• The paper examines the emerging industry of algorithmic auditing and how it is being shaped by new regulatory frameworks like the Digital Services Act (DSA) and Online Safety Act (OSA).
• It explores who is likely to occupy this space, the political and ethical tensions that may arise, and how the mandates of "independent auditing" and "evaluating the societal context of an algorithmic function" will play out in practice.
• The paper warns that the standardizing influence of traditional auditors could erode the regulatory pillars of the DSA and OSA, and that ambitious research ideas for algorithmic auditing may be diluted or compromised.

Plain English Explanation

For almost a decade, researchers have been exploring new ways to audit the inner workings of algorithms and algorithmic systems. This research has now evolved to the point where it is becoming a legal requirement, with regulations like the Digital Services Act (DSA) and Online Safety Act (OSA) establishing frameworks for technology companies and auditors to develop algorithmic auditing practices.

The paper looks at who is likely to take on this role of auditing algorithms, and the potential challenges and tensions that may arise. For example, the idea of "independent auditing" or "evaluating the societal impact of an algorithm" may not be as straightforward as it seems when traditional auditing firms get involved.

The authors warn that the standardized approaches of these traditional auditors could undermine the intent of the new regulations, and that the complex web of contracts, business relationships, and tight timelines could dilute or compromise the ambitious research ideas for algorithmic auditing.

Technical Explanation

The paper systematically reviews the auditing provisions in the DSA and OSA, and compares them to the observations from the emerging algorithmic auditing industry. It explores who is likely to take on the role of algorithmic auditors, and the potential political and ethical tensions that may arise.

The authors analyze how the mandates of "independent auditing" and "evaluating the societal context of an algorithmic function" are likely to play out in practice. They draw attention to the strategies and cultures of traditional auditors, and warn that these could erode the important regulatory pillars established by the DSA and OSA.

The paper also cautions that the ambitious research ideas and technical projects for algorithmic auditing may be compromised by the standardizing influence of traditional auditors, as well as the complex web of contractual arrangements, diverse portfolios, and tight timelines.

Critical Analysis

The paper raises important concerns about the potential pitfalls of integrating algorithmic auditing into existing regulatory frameworks and industry practices. While the authors acknowledge the progress made in algorithmic auditing research, they highlight the risk of this research being diluted or undermined by the involvement of traditional auditors.

One key limitation of the paper is that it does not provide specific examples or case studies to illustrate the tensions and challenges it describes. The authors could have strengthened their arguments by drawing on real-world examples of how algorithmic auditing has been implemented, and the issues that have arisen.

Additionally, the paper could have explored potential solutions or strategies to address the concerns it raises, such as the need for specialized AI audit standards and boards or antitrust measures to ensure algorithmic auditing remains independent.

Conclusion

This paper highlights the tensions and challenges that arise as the emerging field of algorithmic auditing becomes institutionalized through regulatory frameworks like the DSA and OSA. The authors caution that the involvement of traditional auditing firms and industry practices could undermine the intent of these regulations and dilute the ambitious research ideas for algorithmic auditing.

By drawing attention to these issues, the paper encourages readers to think critically about the practical implementation of algorithmic auditing and the need to maintain its integrity and independence. It serves as a valuable contribution to the ongoing discussions around the governance and accountability of algorithmic systems.