The Surveillance State’s First Amendment Problem is No Longer Theoretical

For more than 50 years, the Supreme Court has treated the First Amendment costs of government surveillance as speculation.

In 1972, Laird v. Tatum dismissed a challenge to Army domestic intelligence operations, ruling that a “chilling effect” on protected speech from the mere existence of a surveillance program was too speculative to constitute a cognizable injury.

In 2013, Clapper v. Amnesty International dismissed a challenge to NSA surveillance, ruling that lawyers, journalists and human rights organizations couldn’t prove their communications were actually being collected — even though the government’s surveillance architecture was specifically designed to make that proof impossible.

The result is a closed loop: you can’t sue to challenge surveillance that you can’t prove you’re subject to; you can’t prove you’re subject to it because the government keeps that secret; and the government uses your inability to prove it to defeat standing. The courts call this principled neutrality. Critics call it the structural insulation of the surveillance state from constitutional accountability.

The ground has now shifted beneath those doctrinal assumptions — and the government’s own watchdog has put the evidence on the record.

What the government admits

The Privacy and Civil Liberties Oversight Board, or PCLOB — Congress’ independent civil liberties watchdog inside the executive branch — published a comprehensive Section 702 oversight report in September 2023. The findings are damning, not least because they are self-inflicted.

The Foreign Intelligence Surveillance Act Section 702 is formally limited to targeting non‑U.S. persons located abroad. In 2022, it targeted 246,073 such persons — a 276% increase since 2013.

The intelligence community’s own general counsel has acknowledged that when targeting those persons, the government will “inevitably” collect communications of Americans who communicate with them. The scale of that incidental collection is unknown. The government has refused to estimate it for nine years, claiming it is “infeasible,” despite published cryptographic methodologies that could answer the question without compromising classified sources.

What is known: The FBI conducted up to 3.4 million warrantless back-door searches of Americans’ communications swept into 702 databases in 2021 alone. The Foreign Intelligence Surveillance Court found this constituted “persistent and widespread violations” of governing rules. Among those queried: BLM protesters, donors to a congressional campaign, members of Congress, journalists and crime victims who had contacted the FBI to report crimes against themselves.

In January 2025, a federal-district court became the first in the country to hold that these warrantless backdoor searches ordinarily violate the Fourth Amendment. That ruling is now on appeal to the Second Circuit.

What researchers have established

Laird and Clapper treated “chilling effects” as inherently speculative. For decades, that skepticism had scholarly backing: Frederick Schauer, who wrote the leading theoretical account of chilling effects, acknowledged in 1978 that its behavioral assumptions were “most likely unprovable.”

The Snowden revelations of June 2013 changed that. They created a natural quasi-experiment — a sudden, documented shift in public knowledge of government surveillance — that researchers across disciplines have exploited to measure what courts assumed could never be measured.

Jonathon Penney’s 2016 study in the Berkeley Technology Law Journal found a statistically significant 20% drop in page views on Wikipedia for terrorism-related articles immediately following the Snowden revelations — a drop that persisted over time. Not self-reported fear: documented behavioral change in a wholly legal activity.

Economists Alex Marthews and MIT’s Catherine Tucker found statistically significant declines in Google searches for government-sensitive terms after June 2013 across the United States and 40 trading partners. Tucker, initially skeptical, concluded: “Our article shows that this chill is definitely happening.” Wayne State University’s Elizabeth Stoycheff demonstrated experimentally that surveillance awareness significantly suppresses minority political expression online, amplifying the well-documented “spiral of silence” effect within democratic discourse.

These behavioral findings converge with self-reported evidence from the professional community most directly at risk. PEN America’s 2013 survey of more than 520 American writers — conducted just months after the first Snowden disclosures — found that 85% were worried about government surveillance; 1 in 6 had already avoided writing or speaking about a topic over surveillance fears; 16% had stopped searching online for topics they considered sensitive; and 24% had avoided certain subjects in phone or email communications entirely.

PEN’s 2015 global follow-up of 772 writers in 50 countries found writers in free democracies self-censoring “at levels approaching those seen in non-democratic countries.” Self-reported data has methodological limits, but when stated behavioral changes match up with independently measured ones, the convergence across different methods strengthens confidence that the underlying phenomenon is real. Here they all point the same direction.

The PCLOB acknowledged the consequence directly: “The knowledge that the government can gather this sort of content can have a chilling effect on speech.” Asked whether it had assessed whether Section 702’s incidental collection had produced such effects, the government told the PCLOB it had not — because it had “no reliable way to do so.”

The government has spent years arguing in court that chilling effects are too speculative to be cognizable, while admitting to its own oversight board that it has never once tried to measure whether those effects exist.

A system, not an accident

What these sources together document is a surveillance architecture that suppresses protected First Amendment activity — speech, information-seeking, association and political expression — at population scale, through mechanisms deliberately designed to be opaque. This is done in ways the government has never studied, using authorities insulated from constitutional challenge by a standing doctrine that demands proof that the program’s own secrecy makes impossible to produce. That is not an accident. It is a system.

The social science has moved chilling effects from constitutional speculation to documented behavioral reality. What it has not done — and cannot do, by design — is produce the individualized, “certainly impending” harm that current standing doctrine demands. The courts constructed that demand to keep diffuse, aggregate, structural harms out of Article III. The demand works. That is precisely why legislative remedy — not constitutional litigation — is the operative accountability mechanism.

Section 702 sunsets again in April 2026. The PCLOB recommended in 2023 that Congress require individualized judicial review of U.S. person queries. Congress has reauthorized Section 702 without a meaningful warrant requirement three times since 2013, knowing each time what the program does.

The empirical and legal record is now sufficient to state plainly what Laird and Clapper assumed was speculative: Surveillance at this scale functionally suppresses First Amendment-protected speech, association and inquiry among the American public. The Supreme Court has said that it can’t be proven in court. The social scientists, the professional communities and the government’s own watchdog have now proven it everywhere else.