SEC Predictive Analytics Conflicts Proposal

Overview

The U.S. Securities and Exchange Commission (SEC) issued a proposed rule titled "Conflicts of Interest Associated with the Use of Predictive Data Analytics by Broker-Dealers and Investment Advisers" on July 26, 2023. This proposal, identified as Release No. IA-6353 and File No. S7-12-23, aimed to address the growing concerns regarding the use of advanced technologies, particularly predictive data analytics (PDA) and similar tools, by broker-dealers and investment advisers in their interactions with investors. The core objective of the proposed rules was to prevent firms from placing their own interests ahead of their investors' interests when utilizing these technologies, which have the potential to optimize for firm revenue or other benefits at the expense of clients. The SEC recognized that while PDA could offer significant benefits in market access, efficiency, and returns, its unsupervised or inadequately managed application could exacerbate existing conflicts of interest or create new ones, potentially causing harm to investors on a broader and more pronounced scale due to the scalability of these technologies.

The proposed rules would have broadly imposed a new requirement for firms to evaluate and determine whether their use of certain technologies in investor interactions involved a conflict of interest that resulted in the firm's interests being placed ahead of investors' interests. If such conflicts were identified, firms would have been required to eliminate, or neutralize the effect of, these conflicts. This approach marked a significant shift from traditional disclosure or mitigation strategies, emphasizing proactive elimination or neutralization. The proposal was published in the Federal Register on August 9, 2023, initiating a public comment period. However, as of June 17, 2025, the SEC formally withdrew this and several other proposed rules, stating that it does not intend to issue final rules with respect to these proposals. This withdrawal means that the specific requirements outlined in this proposing release will not come into force, though the underlying concerns about AI and conflicts of interest in financial services remain relevant for future regulatory considerations.

Definitions

The proposed rules introduced several key definitions central to their application. "Predictive Data Analytics" (PDA) referred to technologies that optimize for, predict, guide, forecast, or direct investment-related behaviors or outcomes. This broad definition was intended to capture a wide array of artificial intelligence (AI) and machine learning (ML) applications, including algorithms, models, and other computational methods that process data to generate insights or recommendations influencing investor decisions or firm conduct. The SEC recognized that the increasing sophistication of these models allowed for highly individualized predictions about investors, raising the potential for conflicts where firms might optimize for their own interests over those of their clients.

The proposal also focused on "Covered Firms," which included both "Broker-Dealers" (BDs) and "Investment Advisers" (IAs or RIAs). These entities are regulated under the Securities Exchange Act of 1934 and the Investment Advisers Act of 1940, respectively. The term "conflicts of interest" was central to the proposal, referring to situations where a firm's interests in using PDA-like technologies in investor interactions could lead to outcomes that financially benefit the firm at the expense of investors. This included scenarios where the technology might steer investors towards higher-fee products, generate more trades, or otherwise prioritize firm revenue over client suitability or best interest. The proposed rules aimed to address these conflicts by requiring firms to take steps to eliminate or neutralize their effects, rather than merely disclosing them.

Governance and Institutional Framework

Under the proposed rules, the U.S. Securities and Exchange Commission (SEC) would have served as the primary regulatory authority overseeing compliance. The SEC, leveraging its existing authority under the Securities Exchange Act of 1934 and the Investment Advisers Act of 1940, sought to establish a framework that would mandate specific governance responsibilities for broker-dealers and investment advisers utilizing predictive data analytics. The proposal underscored the SEC's role in protecting investors and maintaining fair, orderly, and efficient markets, extending these principles to the evolving landscape of AI-driven financial services. The proposed rules would have required firms to integrate the identification and management of PDA-related conflicts into their broader compliance and risk management frameworks, thereby making it an integral part of their institutional governance.

Specifically, firms would have been required to establish and maintain written policies and procedures reasonably designed to achieve compliance with the proposed conflicts rules. This would have necessitated internal governance structures capable of overseeing the development, deployment, and ongoing monitoring of predictive data analytics technologies. Such structures would likely have involved cross-functional teams comprising legal, compliance, technology, and business personnel to ensure a holistic approach to conflict identification and neutralization. The proposal also implied a need for robust internal controls and accountability mechanisms to ensure that the use of PDA aligned with investor interests and regulatory expectations. While the proposed rules were ultimately withdrawn, they highlighted the SEC's intent to require firms to take proactive responsibility for the ethical and compliant deployment of AI and PDA within their operations, emphasizing the need for strong internal governance to manage the unique risks posed by these technologies.

Key Focus Areas

The proposed rules primarily focused on identifying and mitigating conflicts of interest arising from the use of predictive data analytics (PDA) in interactions between broker-dealers and investment advisers and their investors. A central tenet was the requirement for firms to "eliminate, or neutralize the effect of," any identified conflicts, a more stringent standard than mere disclosure or mitigation. This meant firms would have needed to conduct a thorough evaluation of their PDA systems to determine if their design or operation could lead to firm interests being prioritized over investor interests, such as optimizing for firm revenue, engagement, or retention at the expense of investor returns or suitability. The scope of "covered technologies" was broad, encompassing any technology that optimizes for, predicts, guides, forecasts, or directs investment-related behaviors or outcomes. This inclusive definition aimed to capture a wide range of AI and machine learning applications, ensuring that the rules remained relevant as technology evolved.

Another key focus area was the comprehensive assessment of all "investor interactions" where covered technologies were used or foreseeably could be used. This included not only direct advisory relationships and recommendations but also broader communications where firms might invite reliance on their expertise. The SEC was particularly concerned about the scalability of these technologies, which could amplify the harm from conflicts of interest across a large investor base rapidly. The proposal also implicitly addressed issues of algorithmic bias and fairness, as conflicts could arise if PDA systems inadvertently or intentionally produced biased recommendations or outcomes that disadvantaged certain investor groups. While not explicitly framed as an anti-bias rule, the requirement to neutralize conflicts would have indirectly compelled firms to consider how their algorithms might create or perpetuate unfair outcomes. The withdrawal of the rules means these specific requirements are not in force, but the underlying regulatory concerns about algorithmic transparency, fairness, and conflict management in financial AI remain pertinent for firms operating in this space.

Implementation Framework

Had the proposed rules been adopted, broker-dealers and investment advisers would have been required to establish a robust implementation framework to comply with the new mandates concerning predictive data analytics (PDA). This framework would have necessitated a fundamental shift in how firms approached the integration of AI and machine learning into their investor interactions. Central to this would have been the development and maintenance of written policies and procedures specifically designed to identify, eliminate, or neutralize conflicts of interest associated with their use of covered technologies. These policies would need to be tailored to the specific technologies employed by each firm and the unique ways in which those technologies interacted with investors, ensuring that the firm's interests were not prioritized over those of its clients.

The implementation framework would also have required firms to conduct regular and comprehensive evaluations of their PDA systems. This would involve assessing the design, development, testing, and deployment phases of these technologies to proactively identify potential conflicts. Firms would have needed to establish clear processes for documenting these evaluations, including the methodologies used, the conflicts identified, and the specific actions taken to eliminate or neutralize them. Furthermore, the proposed rules would have necessitated amendments to existing compliance programs to integrate these new requirements, potentially including staff training on the risks of PDA, the new conflict identification standards, and the firm's specific policies and procedures. The goal was to embed conflict management for AI-driven tools deeply within the firm's operational and compliance infrastructure, moving beyond traditional disclosure models to a proactive elimination or neutralization approach. Despite the withdrawal of the proposed rules, the principles of responsible AI governance and conflict management remain critical considerations for financial firms.

Monitoring and Evaluation

The proposed rules would have imposed significant obligations on broker-dealers and investment advisers regarding the ongoing monitoring and evaluation of their predictive data analytics (PDA) systems. Firms would have been required to implement continuous monitoring mechanisms to ensure that the technologies, once deployed, did not inadvertently develop new conflicts of interest or fail to effectively neutralize existing ones. This would have involved regular reviews of the performance and outputs of PDA models, assessing whether their recommendations or guidance continued to align with investor interests and did not prioritize firm benefits. Such monitoring would likely have included periodic audits of data inputs, algorithmic logic, and output interpretations to detect any drift or emergent biases that could lead to conflicts.

Furthermore, the evaluation framework would have extended to the effectiveness of the policies and procedures designed to manage these conflicts. Firms would have needed to regularly assess whether their established controls were adequate and functioning as intended, making adjustments as necessary to adapt to changes in technology, market conditions, or investor behavior. This iterative process of monitoring and evaluation would have been crucial for maintaining compliance with the proposed rules and ensuring ongoing investor protection. Record-keeping requirements, which were also part of the proposal, would have supported this by mandating documentation of all monitoring activities, evaluation findings, and any remedial actions taken. While these specific rules have been withdrawn, the importance of robust monitoring and evaluation practices for any AI-driven system in financial services remains a best practice for risk management and ethical conduct.

Penalties, Liability, and Appeals

Had the proposed rules on conflicts of interest associated with predictive data analytics been adopted, non-compliance would have subjected broker-dealers and investment advisers to the existing enforcement mechanisms and penalties under the U.S. federal securities laws. The Securities and Exchange Commission (SEC) possesses broad authority to investigate and prosecute violations of the Securities Exchange Act of 1934 and the Investment Advisers Act of 1940, which would have formed the statutory basis for these new rules. Penalties for non-compliance could have ranged from cease-and-desist orders, censures, and fines to disgorgement of ill-gotten gains and suspension or revocation of registration. The severity of the penalty would have depended on factors such as the nature and extent of the violation, the harm to investors, and the firm's history of compliance. The SEC's enforcement actions are often public, serving both as a deterrent and as a means to inform the industry about regulatory expectations.

Firms or individuals facing SEC enforcement actions would have had various avenues for appeal, typically through administrative proceedings before an SEC administrative law judge, followed by review by the Commission itself, and ultimately, judicial review in federal courts. The legal liability arising from breaches of the proposed rules could also have extended to private rights of action, where investors harmed by a firm's failure to eliminate or neutralize conflicts of interest could potentially sue for damages. This dual layer of public enforcement and private litigation underscores the significant legal and financial risks associated with non-compliance in the financial sector. However, since the proposed rules were formally withdrawn by the SEC as of June 17, 2025, these specific penalties and liabilities tied directly to this proposal will not come into effect. Nevertheless, firms remain subject to existing regulations concerning conflicts of interest, fiduciary duties, and investor protection, which may still apply to the use of AI and PDA.

Relationship to Other Instruments

The proposed rules on conflicts of interest associated with predictive data analytics (PDA) were intended to build upon and complement existing regulatory instruments governing broker-dealers and investment advisers. Specifically, they would have interacted significantly with the established fiduciary duties owed by investment advisers to their clients and the best interest obligations imposed on broker-dealers under Regulation Best Interest (Reg BI). While existing rules require firms to address conflicts of interest, the proposed rules aimed to provide a more specific and stringent framework for conflicts arising from the unique characteristics of PDA, moving beyond mere disclosure to a requirement of elimination or neutralization. This would have meant that firms could not simply disclose a PDA-driven conflict but would need to actively remove or mitigate its adverse effects on investors, potentially requiring a re-evaluation of how these technologies are designed and deployed.

Furthermore, the proposed rules would have intersected with other SEC regulations concerning record-keeping, compliance policies and procedures, and supervision. Firms would have been required to amend their existing compliance programs to incorporate the new requirements, ensuring that their use of PDA was consistent with all applicable securities laws and regulations. The proposal also referenced the Securities Exchange Act of 1934 and the Investment Advisers Act of 1940 as the foundational legal authorities for the new rules, indicating that it was an extension of the SEC's long-standing mandate to protect investors. While the proposed rules themselves have been withdrawn, the SEC's action highlighted a recognition that existing frameworks, while robust, may need specific augmentation to adequately address the novel challenges posed by advanced AI technologies in financial markets. The withdrawal suggests that the SEC may consider alternative approaches or rely on evolving interpretations of existing rules to address these concerns in the future.

International Alignment

The U.S. Securities and Exchange Commission's proposed rules concerning conflicts of interest associated with predictive data analytics by broker-dealers and investment advisers, while a domestic initiative, reflected a broader global trend among financial regulators to address the implications of artificial intelligence (AI) and advanced analytics in financial services. Jurisdictions worldwide, including the European Union, the United Kingdom, and various Asian financial hubs, are actively exploring or implementing regulatory frameworks to manage the risks and opportunities presented by AI in finance. Common themes across these international discussions include concerns over algorithmic bias, data privacy, explainability, cybersecurity, and, crucially, conflicts of interest, particularly where AI systems might optimize for commercial outcomes at the expense of consumer protection.

Although the SEC's proposal was specific to the U.S. securities market, its underlying principles—such as the need for firms to identify and mitigate conflicts arising from AI, ensure fairness, and maintain robust governance—resonate with international regulatory dialogues. For instance, the European Union's proposed AI Act, while broader in scope, includes provisions for high-risk AI systems, which could encompass certain financial applications, mandating risk management systems, data governance, and human oversight. Similarly, other international bodies and national regulators have issued guidance or initiated consultations on AI ethics and governance in finance. While the SEC's specific proposed rules have been withdrawn, the global regulatory landscape continues to evolve, with ongoing efforts to establish consistent principles for responsible AI development and deployment in financial services to ensure market integrity and investor protection across borders. The withdrawal does not diminish the international relevance of the issues the proposal sought to address.

Implementation Timeline

Compliance Checklist

Sources and References