Letter to the SEC regarding First Amendment limits on jurisdiction

A PDF version of this filing is available here.

Dear Chair Atkins and Commissioner Peirce,

This letter addresses a threshold issue for the Commission’s approach to digital assets: the First Amendment limits on SEC jurisdiction over software development, publication, and neutral web and blockchain infrastructure.

Those limits should not be left to emerge only as affirmative defenses to Commission enforcement actions. They can and should be articulated ex ante by the Commission itself. Without clear guidance, existing broker-dealer and exchange concepts—already broad and underdettermined—risk being applied by proximity to activities that are, in substance, the publication of software and information. In the context of modern computing, that ambiguity does not merely create uncertainty; it risks chilling the constitutionally protected speech that is at the foundation of these innovative technologies.¹

The Commission is well positioned to address this directly by clarifying the boundary between publication and regulated conduct and by ensuring that its investigative and enforcement processes respect that line. Coin Center urges the Commission to tackle these difficult topics through an interpretive ruling on the Constitutional limits to its authority. To assist in that work, we are attaching our newly published report on the application of First Amendment jurisprudence to crypto software development alongside this letter.

The Commission’s recent Staff Statement² concerning broker-dealer registration for certain user interfaces already points in the right direction. It distinguishes between the mere act of making software publicly available and activities involving control, discretion, or participation in user transactions. We also agree with Commissioner Peirce’s associated commentary that the staff statement, while helpful, should be followed by “a more permanent regulatory approach.”³ The distinctions at play touch on the most permanent passages of American law, the Constitution, and our rights as Americans to speak and publish free from prior restraint and without compulsion to express messages preferred by the government but not the speaker.

As suggested by Commissioner Peirce, there is now an opportunity to make these principles explicit. Congress is already moving in a similar direction, with current legislative proposals urging clearer limits on registration authority consistent with the First Amendment. But the Commission need not wait for legislation. The constitutional boundary exists irrespective of any new statutory authority, and Commission-level guidance articulating that boundary would provide durable clarity to both market participants and staff.

There are two core constitutional principles ripe for clarification. First, that the First Amendment protects the publication of software, data, technical designs, and neutral communications tools. Second, that registration requirements are only constitutionally permissible as prior restraints where a person takes another’s financial affairs in hand in the traditional sense of an agent, principal, custodian, advisor, or manager who is exercising client-specific discretionary authority.⁴

The first principle can be dealt with swiftly. The Commission can explicitly reject a recurring doctrinal error in this area. Some lower courts have suggested that software may receive diminished First Amendment protection because it is “functional” or capable of producing real-world effects. That approach has no basis in Supreme Court precedent. The Court has repeatedly held that the creation and dissemination of information is speech, even when that information is technical, commercially valuable, or used to guide conduct.⁵ The Commission should make clear that, following binding Court precedent, it does not treat software as a lesser category of speech and does not treat the fact that software can be used to perform actions—or to facilitate activities that were historically intermediated—as a basis for diminishing First Amendment protection or expanding its jurisdiction.

With that constitutional baseline in place, the relevant question is not whether software is useful or capable of facilitating transactions, but whether a person has moved beyond software publication into a role long understood to be subject to regulation. The Commission can provide durable clarity by articulating this boundary in concrete terms.

For this, we propose an analytical framework, structured as a sequence rather than an open-ended balancing test, composed of three parts:

First, does the person have custody of, or unilateral control over, the securities or funds of others? If the answer is yes, the First Amendment limit is much weaker and ordinary regulation may apply because custody or unilateral control is a strong marker of conduct rather than publication, as a person who can seize, redirect, hold, or dispose of another’s assets is in a very different posture from a publisher or infrastructure provider.

Second, even without custody, does the person retain ongoing discretionary power over how securities transactions are facilitated after or alongside publication? If not, and the person is publishing software or providing tools of general use, the activity is firmly protected by the First Amendment. Mere publication of tools that investors choose to employ to perform securities transactions does not become regulated conduct just because the tools are useful for conducting transactions that once would have required an intermediary.⁶ This includes publishing new versions of software, recommending updates, or attempting to persuade users to adopt improved or modified tools. Those activities remain speech, even where they are effective in influencing user behavior.⁷

The relevant question is whether the person retains the ability to affect user transactions or positions without fresh user assent. The hard cases arise only where a person maintains a continuing role that allows them to alter outcomes, routes, or conditions of transactions in a way that goes beyond user-directed action.

Third, where some ongoing discretionary role is retained, is that role merely ministerial or communicative, or is it part of an agency, fiduciary, or client-specific advisory relationship of the sort described in Lowe v. SEC?⁸

Not all discretion is relevant in this context. Many systems involve automated processes, parameter-setting, routing logic, or the aggregation and presentation of information. Those features may influence user decisions, but they do not, by themselves, constitute the exercise of professional judgment on behalf of another.

The relevant distinction is whether the asserted discretion is exercised for the user in light of that user’s particular circumstances, or whether it is exercised for general system design, maintenance, or communication. Where the activity consists of repackaging, organizing, or presenting public information, or providing tools that users independently employ to make their own decisions, it remains ministerial or communicative—even where the tools are highly sophisticated or effective.

This is consistent with the distinction recognized in Lowe, where the Court drew a line between generalized publications available to the public and conduct in which a professional takes a client’s affairs personally in hand. Modern financial tools such as data terminals and analytics platforms may dramatically reduce information asymmetries and enable rapid, informed decision-making, but their developers do not thereby become fiduciaries or intermediaries. The same principle applies here.

Where discretion operates only at this general, system-wide level, the activity remains presumptively protected. Where it instead involves delegated, client-specific judgment or authority over transactions, regulation is more plausible. The fact that a tool makes decision-making easier, faster, or more informed does not convert its publisher into a regulated intermediary.

Using this framework, an interpretive guidance should clarify that certain activities be treated as presumptively ministerial or communicative. These activities can be separated into four categories, all of which inherently contain these properties.

The first category are those activities that consist of deterministic integrity and validation functions, such as: validating digital signatures, checking transaction formatting, verifying data integrity, and enforcing deterministic protocol rules. These various activities are all mechanical checks and not professional financial judgment, and should therefore be treated as ministerial.

The second category are those activities that consist of consensus ordering and network processing, as these are incidents of network operation and communication, not agency. These sorts of activities include ordering transaction data within an open proof-of-work or proof-of-stake mechanism, block construction or validation according to open protocol rules, and routine propagation or relaying of transaction messages or blocks.

The third are those that consist of repackaging or reorganizing public information, which remain protected so long as the service does not provide individualized financial advice in light of a client’s specific objectives, risk tolerance, or circumstances per Lowe. These activities include displaying on-chain data in more usable form, quoting observed network fees or likely fee estimates for settlement or finality, presenting liquidity, pricing, or blockchain state information derived from public sources, and assisting users in viewing and interacting with a blockchain.

The fourth and last category are those activities that consist of publishing or deploying general transaction tools. In these cases, the user initiates, chooses, and signs the transaction, and the developer or publisher does not take over the user’s financial affairs. This includes publishing or deploying open-source or source-available automated market makers (AMMs), matching engines, wallet software, and routing or transaction-construction tools.⁹

Importantly, this framework would not encompass the edge cases of control and client-specific professional conduct that would move beyond First Amendment protections and into regulable conduct. Some indicators that an activity may be regulable conduct include: custody or unilateral asset control, authority to execute, alter, or block transactions on behalf of users, ongoing discretionary management of user assets or strategies, individualized financial advice tied to a client’s circumstances, an agency or fiduciary relationship of trust and reliance, or bespoke services where the provider is effectively taking a client’s affairs personally in hand.

It is also important to distinguish between permissible regulation of conduct that incidentally burdens speech and impermissible regulation that compels the redesign of speech itself. The Commission may regulate activities such as custody, execution, or agency relationships, even where doing so has incidental effects on how firms communicate. But where a theory of liability depends on requiring a developer or publisher to alter the design of software—such as by introducing user identification, monitoring, or intermediation functions that the software would not otherwise contain—the burden is no longer incidental. It is a direct compulsion of speech and expressive design.

The Treasury Department’s 2024 non-custodial broker-reporting rulemaking illustrates the problem.¹⁰ That proposal would have extended reporting obligations to software developers and other non-intermediaries who neither effected transactions as agents nor maintained customer relationships in any traditional sense. Because those actors did not possess the required information, compliance would have required redesigning their software to collect and report user data. In effect, the rule attempted to solve the absence of a traditional intermediary by forcing publishers of self-directed tools to become intermediaries. That move is not a regulation of conduct with incidental effects on speech; it is a compelled transformation of speech into regulated activity.

The same concern should guide the Commission’s own approach. Where regulation depends on treating the publisher of a tool as the “missing middleman,” and thereby requiring changes to the content or architecture of the software itself, the First Amendment is directly implicated.

Additionally, for this framework to be most effective, we recommend that the Commission build internal procedures that require First Amendment considerations before investigations escalate. This is important because constitutional limits are less meaningful if they appear only as after-the-fact defenses in court, and the Commission would be better equipped to protect speech, improve staff discipline, and create a reviewable administrative record. This is especially important as the Commission may be faced with cases attempting to litigate the line between software development and regulable conduct, considering the rapid evolution of the technology. This approach would also reduce chilling effects on software development and web services, develop a clearer record of agency reasoning for the courts, avoid turning constitutional boundaries into private litigation costs, and does not require Congress or sweeping substantive exemptions.

For these reasons, Coin Center proposes implementing a checklist prior to a Wells Notice or formal action, adhering to recordkeeping and transparency, and conducting Commission-level review for edge cases.

Specifically, before opening a formal investigation premised on broker, dealer, exchange, adviser, or similar theories against a software developer, publisher, or web service provider, staff should document: whether the target had custody of securities or funds of others; whether the target had unilateral power to alter, route, block, or execute transactions; whether the target exercised ongoing discretion over transaction facilitation; whether any such discretion was merely ministerial or communicative; whether the target had an agency, fiduciary, or client-specific advisory relationship with users; whether the target used user-specific information to exercise financial judgment on behalf of users; whether the theory of liability rests on publication / infrastructure activity alone; and whether enforcement would burden protected speech, and if so whether the burden is merely incidental to regulation of conduct rather than direct regulation of publishing as such.

Staff should also create a written record of this analysis before Wells notice or comparable escalatory action. That record need not be public immediately, but it should exist so that the Commission, reviewing courts, and defendants can understand how the Commission reasoned under its own policy. And where staff believes action is warranted despite substantial speech concerns, escalation should require Commission-level or senior-level approval. This would ensure novel or aggressive theories receive scrutiny before reputationally damaging steps are taken.

To summarize, we recommend that the Commission issue formal guidance stating that the First Amendment limits SEC jurisdiction over neutral software publication and neutral web and blockchain infrastructure, and that ordinary securities regulation becomes more plausible only where agency, custody, client-specific discretion, or similar professional conduct is present. We also recommend that the Commission state that the specific categories listed above are presumptively ministerial or communicative and therefore presumptively protected absent additional facts showing client-specific professional conduct. And lastly, that the Commission adopt an internal checklist and written-review process for investigations and Wells notices implicating software publication, web services, or blockchain infrastructure.

This approach would also be consistent with the Commission’s own longstanding recognition of the constitutional limits on its authority. As Commissioner Roberta Karmel explained in 1979, there is a “tension between the First Amendment and federal securities laws” that should drive the Commission to adopt “rules in a less restrictive form.”¹¹ That principle applies with equal force here. Where regulatory approaches risk burdening protected publication—particularly by treating software developers as if they were intermediaries—the Commission has both the authority and the obligation to adopt narrower, constitutionally sound interpretations explicitly to the public and internally through enforcement and investigatory procedures and safeguards.

Sincerely,

Peter Van Valkenburgh
Executive Director, Coin Center

Notes