The Supreme Court’s test for determining whether speech-related conditions on federal grants violate the First Amendment seems straightforward, at least initially. The Court has drawn a distinction “between conditions that define the limits of the government spending program—those that specify the activities Congress wants to subsidize—and conditions that seek to leverage funding to regulate speech outside the contours of the program itself.”

So stated, this distinction seems to balance two important considerations: (1) the widely accepted understanding that the government gets to engage in its own speech in support of the favored points of view of elected officials; and (2) the worry that the government will weaponize reliance on federal grants to pressure those outside the government into skewing their own private speech toward those favored points of view.

Government speech—like all other forms of government action in the modern era—often takes place through privatization or subsidization. And if the government is to pay others to deliver its message, it must be able to require that those others hew to its preferred script. When the government provides grants to outside entities to engage in a pro-immunization advertising campaign, it can surely prohibit those entities from spending the money on promoting anti-vax conspiracy theories.

Yet because people and institutions rely heavily on government funds, the threat to withhold those funds can create strong pressure to hew to the official line, even outside of the programs specifically paid for by the government. The inside-the-program/outside-the-program distinction seems to respond to this fear by barring the government from basing funding decisions on external speech.

On closer inspection, though, the Court’s crucial distinction seems analytically quite flimsy. There is little doubt the government could lawfully create two separate programs, one paying institutions to conduct medical research and the other paying institutions to promote the war on Gaza. And there is no general prohibition on tying together multiple lawful objectives in one funding program. Nor could there be such a general prohibition if government is to operate effectively. Rather, “both common sense and a basic understanding of politics reveal that federal funding legislation (like any legislation) typically has multiple purposes.”

Altogether, this means that virtually any condition could be formally described as “defin[ing] the limits of the government spending program.” If the government conditions the receipt of federal medical research grants on an institution’s support of Israel’s war on Gaza, the grants could formally be described as having two purposes: promoting medical research and encouraging speech in support of the war.

Yet allowing the government to tie these conditions together like that would seem to skew the balance the Court has attempted to draw in its unconstitutional conditions jurisprudence. It would give the government greater ability to engage in its own speech, but only at the expense of dramatically expanding the government’s power to skew private speech.

The Supreme Court has recognized the problem. In Agency for International Development v. U.S. Alliance for Open Society International, Chief Justice Roberts’ opinion for the Court acknowledged that “the definition of a particular program can always be manipulated to subsume the challenged condition.” Roberts continued by reaffirming the Court’s earlier statement in Legal Services Corporation v. Velazquez that “Congress cannot recast a condition on funding as a mere definition of its program in every case, lest the First Amendment be reduced to a simple semantic exercise.”

The Court attempted to heed that caution in Agency for International Development itself. There, it invalidated a statutory provision that required certain recipients of HIV prevention and treatment funds to “have a policy explicitly opposing prostitution and sex trafficking.” The Court took for granted that Congress acted lawfully in enacting a separate provision that prohibited using the HIV prevention and treatment funds “to promote or advocate the legalization or practice of prostitution or sex trafficking,” but it held that the requirement to have an anti-prostitution and anti-trafficking policy “demand[ed] that funding recipients adopt—as their own—the Government’s view on an issue of public concern” and thus “by its very nature affect[ed] protected conduct outside the scope of the federally funded program” (internal quotation marks omitted).

But the caution that Congress can’t avoid the unconstitutional conditions rule by definitional fiat does not give us a rule for determining what is properly considered a definition of the scope of the program versus what is properly considered an effort to regulate speech outside of the program. And Chief Justice Roberts admitted this in Agency for International Development. He said the line between program definition and external regulation “is hardly clear,” and later in the same opinion reiterated that the distinction “is not always self-evident.”

These are understatements. The distinction between program definition and external regulation itself cannot analytically provide a principle for deciding what sorts of rules belong in which category. At most, it can tell us that there have to be two categories, so the government cannot collapse every condition into the definition of the program. But it cannot tell us how to decide what counts as regulation of external speech. To do that doctrinally requires articulating some sort of mediating principle or principles, to be incorporated in one or more doctrinal subrules that will be employed when courts apply the Agency for International Development rule.

Let me identify one possible subrule (though there are others one might derive from the cases as well): The Supreme Court’s cases can be read to require the government to structure its grants so they give grantees a meaningful opportunity to engage in their own independent expression on the relevant topics, even if doing so comes at some risk of undermining the message the government intends to send through its program.

This possible subrule emerges from Chief Justice Roberts’ discussion, in his opinion in Agency for International Development, of the Court’s prior unconstitutional conditions cases. The Court had previously upheld a prohibition on lobbying by 501(c)(3) organizations in Regan v. Taxation With Representation, as well as a prohibition on promoting abortion by projects that received federal family planning funds in Rust v. Sullivan.

But Agency for International Development stressed that in both cases Congress had given organizations meaningful opportunities to segregate their federally subsidized activities (in which the government could limit their speech) from their other activities (in which the organizations could speak as they wished). Thus, in Regan a tax-exempt organization retained the ability to lobby “[b]y returning to a ‘dual structure’ it had used in the past—separately incorporating as a § 501(c)(3) organization and § 501(c)(4) organization.” Agency for International Development (quoting Regan). Under such a structure, which Regan noted “was not ‘unduly burdensome,’” the organization “could continue to claim § 501(c)(3) status for its nonlobbying activities, while attempting to influence legislation in its § 501(c)(4) capacity with separate funds.” Id. (quoting Regan). Similarly, Agency for International Development explained that in Rust “the Court stressed that ‘Title X expressly distinguishes between a Title X grantee and a Title X project,’” and that the speech limitations “governed only the scope of the grantee’s Title X projects, leaving it ‘unfettered in its other activities.’”

The Agency for International Development Court contrasted these cases with the law held unconstitutional in FCC v. League of Women Voters. That law barred public broadcasters that received federal assistance from any “editorializing, including with private funds.” Crucially, and unlike in Regan and Rust, “the law provided no way for a station to limit its use of federal funds to noneditorializing activities, while using private funds ‘to make known its views on matters of public importance.’” Agency for International Development (quoting League of Women Voters).

It is fair to read this discussion as requiring the government to give grantees the opportunity (and a “not ‘unduly burdensome’” opportunity) to segregate federally funded activities from activities supported by other funds, with the government’s speech restrictions applying only to the former category. Such a requirement necessarily reflects a balance. Federal funding of some of an organization’s activities may free up money for the organization to spend on its other activities, even if the two sets of activities are formally separate. And federally funded organizations may find a way to use the freed-up money in their private accounts to speak in ways that run counter to the government’s preferred message. Justice Scalia made this point in his dissent in Agency for International Development: “Money is fungible. The economic reality is that when NGOs can conduct their AIDS work on the Government's dime, they can expend greater resources on policies that undercut the Leadership Act.” But the Court was willing to accept some inefficiency in the government’s ability to spread its own message in the interest of ensuring that the government could not use its economic power and deep financial entwinement with private grantees as a club to coerce the speech of those grantees.

Under this principle, several demands the Trump Administration made of Harvard would likely be unconstitutional even if there were a statutory basis for them. The demand that Harvard promote ideological diversity in each program and field, for example, clearly seeks to regulate speech without giving the university any opportunity to segregate any department or field from federal funds. Similarly, the demand that Harvard “immediately shutter all diversity, equity, and inclusion (DEI) programs, offices, committees, positions, and initiatives, under whatever name, and stop all DEI-based policies,” would require the university to abandon efforts to instruct faculty and administrators about how bias instantiates—efforts that are pure speech—without, again, providing any opportunity to engage in these programs with private funds only.

It is fair to ask, at this point, whether the subrule I have identified would call into doubt the constitutionality of the Civil Rights Restoration Act of 1987 (CRRA). That statute (actually adopted in 1988, despite its name) clarified that Title VI of the Civil Rights Act, Title IX of the Education Amendments, and Section 504 of the Rehabilitation act prohibit discrimination in “all of the operations of” an institution, including “a college, university, or other postsecondary institution, or a public system of higher education ... any part of which is extended Federal financial assistance.”

The CRRA thus covers an entire university, any part of which receives federal grants. But I would submit that the CRRA does not violate the principle that the government must give grantees a meaningful opportunity to segregate portions of their operations from federal funding so that they may speak as they wish in their non-federally-funded activities. Title VI, Title IX, and Section 504, properly read, do not regulate speech. They prohibit discriminatory conduct. Sometimes that conduct takes the form of discriminatory harassment, but (in part to protect free speech) the standard for finding unlawful harassment is high: In the education context, the harassment must be “so severe, pervasive, and objectively offensive that it can be said to deprive the victims of access to the educational opportunities or benefits provided by the school.” As one commentator explained, “[p]roperly understood, hostile environment law under [these statutes] has a limited scope and should not be interpreted to encompass various protected expressions.”

So long as the substantive requirements of the civil rights statutes are kept within their proper bounds—a point the Trump Administration has not heeded in its assault on universities—they do not even implicate the First Amendment doctrine of Agency for International Development. Because Congress could directly require all universities to avoid discriminatory harassment, without conditioning that requirement on federal funds, the limitations on conditional funding are not relevant.

Notably, although the Trump Administration is relying on Title VI as a reason for canceling federal grants for Columbia, Harvard, and other institutions, there are strong arguments both that those universities did not violate Title VI and that, to the contrary, the government has violated the procedures Congress specified for enforcing the statute. Although the Trump Administration’s actions plausibly violate the Agency for International Development principle based on my analysis above, nothing in that analysis calls into question the constitutionality of the civil rights statutes when they are properly applied according to their terms.