Can a society without law and order trust security keys held by a third party?

Cryptography is often marketed as certainty: a branch of mathematics that does not bend to politics, panic, or the persuasion of retaliation, unemployment or canceled contracts. An encryption key is either secure or it is not. A signature is either valid or it is not. In that narrow, technical sense, keys are trustworthy—because they are not moral actors.

But modern “security keys” rarely live in a vacuum. Increasingly, they are held, mediated, synchronized, recovered, or attested by someone else: a platform vendor, a password manager, a hardware manufacturer, an identity provider, or a cloud service. Even passkeys—marketed as the end of passwords—typically depend on ecosystems that include device operating systems, backup and sync services, and account recovery mechanisms. This is not inherently sinister; it is how usable security is built at scale. Yet it forces a civic question that technology alone cannot answer: Can a society that cannot reliably impose consequences on those who exercise unchecked or coercive power trust security keys that are, in practice, held or influenced by a third party?

This is not a fresh question. But those of us who have cautioned about these matters have often been dismissed as alarmists or conspiracy theorists.

In the 1990s, the United States government promoted the “Clipper Chip” as part of the Escrowed Encryption Standard: strong encryption for communications paired with a built-in key escrow mechanism so law enforcement could decrypt when “lawfully authorized” (National Institute of Standards and Technology [NIST], 1994). From the start, the fight was about more than circuitry. Key escrow was a claim that the state should have a durable, engineered path to plaintext. Critics heard something else: that the state wanted encryption that citizens could use only so long as the state maintained privileged access.

Technical critique arrived quickly. Matt Blaze’s 1994 analysis identified serious weaknesses in the escrow protocol design, sharpening public skepticism about whether “lawful access” could be engineered without creating new failure modes and new powers (Blaze, 1994). NIST responded publicly, defending the technology and emphasizing that it was voluntary—an early illustration of a recurring pattern: when public trust is low, “voluntary” often reads as “voluntary until it isn’t” (NIST, 1994).

Then came another object lesson: Lavabit. In 2013, Lavabit became the center of a high-profile legal conflict after U.S. investigators sought the service’s TLS/SSL private key—an asset that could, by design, expose the security of the entire service, not merely one user (Zetter, 2013). Court records in the Fourth Circuit describe a sequence of orders—under the Pen/Trap statute and the Stored Communications Act—pressuring Lavabit to provide keys (United States v. Lavabit, LLC, 2014). Lavabit’s founder argued that surrendering those keys would put all customers at risk, and the company ultimately shut down rather than continue operating under those demands (Zetter, 2013). Whether one sees that as principled resistance or stubborn obstruction, the civic message was unmistakable: when the state wants access badly enough, the technical and legal fight can shift from “give us the target” to “give us the infrastructure.”

In the wake of the San Bernardino attack, these tensions escalated into the Apple–FBI dispute over a locked iPhone, with the government seeking to compel Apple to assist in bypassing security protections (Cook, 2016; Gibbs, 2016). Around that period, Senator John Cornyn publicly participated in the “going dark” framing in Senate Judiciary discussions: pressing FBI Director James Comey on whether encryption was “part of terrorist tradecraft,” whether end-to-end encryption could make communications unreadable even with a court order, and whether Congress should “do something about this” rather than rely on voluntary industry cooperation (U.S. Senate Committee on the Judiciary, 2015; Sforza, 2015). Again, the rhetorical contest was about more than a phone. It was about who gets the last word: the citizen holding a key, or the institutions that can demand compliance.

Against that history, the present question becomes sharper—not because encryption is weaker now, but because the social conditions around encryption can be weaker.

Keys are claims about power, not just secrets

A private key is an exclusion mechanism. It says: only the holder can decrypt; only the holder can sign; only the holder can authenticate. Passkeys—a FIDO/WebAuthn approach intended to replace passwords—strengthen this promise by making authentication phishing-resistant and by avoiding server-side password databases (FIDO Alliance, n.d.; World Wide Web Consortium [W3C], 2021). In U.S. government guidance, phishing-resistant authenticators are treated as central to modern authentication practice (Temoshok et al., 2025).

So far, so good: fewer passwords to steal, less credential stuffing, and a smaller blast radius when databases leak.

But a key only functions as freedom if the holder can refuse. The moment refusal becomes impossible—through seizure, compulsion, intimidation, or consequence-free violence—the private key stops being “private” in the way that matters. The math still works. The social promise does not.

That is why the question of “law and order” belongs in the encryption debate. “Law and order,” at its best, does not mean harshness. It means predictable restraint: rules that bind the powerful as well as ordinary people, procedures that are followed, investigations that are credible, and consequences that are real when wrongdoing occurs. When those conditions fail, trust does not simply decline—it becomes irrational to maintain.

When bodily impunity feels plausible, digital trust becomes fragile

Incidents like the recent killing of Alex Pretti in Minneapolis matter for the encryption debate because they shape what citizens believe about accountability—especially when video evidence, competing official narratives, and jurisdictional complexity collide.

ABC News compiled a minute-by-minute timeline based on multiple verified videos of the encounter with federal officers in Minneapolis on January 24, 2026, reporting that the videos did not appear to support certain official assertions about Pretti’s behavior during the incident (ABC News, 2026). Separate reporting described sworn witness accounts and a physician’s affidavit alleging that Pretti was pinned on the ground and suffered multiple gunshot wounds, with most wounds described as being in his back (The Guardian, 2026). Meanwhile, local reporting has emphasized how difficult it can be to prosecute federal officers in state court, given doctrines and defenses that can narrow state authority even when state officials seek accountability (Day, 2026).

This article does not pretend to adjudicate guilt from afar. The point here is civic, not forensic. When a society watches an encounter that appears (to many viewers) to involve overwhelming control of a suspect and then lethal force (read "extra-judicial execution")—followed by an accountability pathway that looks structurally obstructed— people absorb a lesson about where power ultimately resides. And once a society internalizes that lesson, it becomes rational to ask: if coercive power can be applied to bodies without credible consequence, why would anyone assume it will not be applied—directly or indirectly—to keys?

The third-party reality: most keys now live inside institutions

When people hear “third party” in cryptography, they often think of overt key escrow or mandated backdoors. That is only one—and the most politically radioactive—form of third-party involvement. In everyday life, third parties sit in the system in subtler ways:

Platform vendors maintain secure enclaves, keystores, and credential managers where keys reside.
Cloud services sync credentials across devices.
Identity providers broker authentication decisions and session issuance.
Recovery workflows become human and bureaucratic override channels.
Hardware makers control firmware updates and attestation behavior.

None of this necessarily means a third party literally possesses your private key. But it often means a third party controls the conditions under which you can use your key—and the processes by which someone else can attempt to make you use it.

In a society with strong rule-of-law norms, people accept this arrangement because there are guardrails: warrants, judicial review, public scrutiny, meaningful remedies, and credible consequences for abuse. In a society where accountability is perceived as optional for those who wield force, those guardrails start to look like theater. The “third party” stops looking like a reliability layer and starts looking like a leverage point.

End-to-end encryption does not eliminate coercion; it reallocates it

Platform providers rightly emphasize end-to-end encrypted syncing for certain keychain and credential systems. Apple, for example, describes iCloud Keychain as end-to-end encrypted such that even Apple cannot read the contents while in transit or storage (Apple, 2024). That matters. It reduces exposure to mass compromise, certain insider risks, and compelled disclosure demands aimed at the provider.

But it does not answer the civic question. End-to-end encryption can protect data from the cloud provider; it cannot protect data from the fact that devices live in pockets, homes, workplaces, and traffic stops—and that the state has unique powers over those spaces. Nor does it eliminate the reality that third parties still govern critical parts of the lifecycle: device security policy, credential syncing availability, account recovery, and compliance posture when served with orders.

Lavabit illustrates the “infrastructure squeeze” problem: when the state cannot easily isolate a single target, pressure can shift to shared technical choke points, like service keys (United States v. Lavabit, LLC, 2014; Zetter, 2013). Clipper illustrates the “built-in access” temptation: engineer lawful access into the design itself (NIST, 1994). San Bernardino illustrates the “compelled assistance” strategy: if you cannot break the device, compel the maker to help (Cook, 2016; Gibbs, 2016). In each case, the social question is the same: what restrains power when power wants access?

The passkey paradox: better security can increase the value of coercion

Passwords are weak, but they diffuse power. A password can be forgotten, mistyped, reset, or plausibly denied as compromised. Passkeys reduce those failure modes. They also reduce a user’s ability to resist coercion through friction.

Passkeys are designed to be fast and low-friction: approve with a biometric or device PIN and you are in (FIDO Alliance, n.d.; W3C, 2021). From a user-experience perspective, that is the point. From a civil-liberties perspective, it changes the texture of compulsion. If authentication becomes as simple as placing a device in front of someone’s face or pressing their finger to a sensor, then the boundary between consent and coercion can become thinner in practice, even if it remains clear in law.

This is not an argument against passkeys. It is an argument that authentication technology cannot substitute for institutional restraint. And there is no restraint if there are not consequences for state actors who attempt to ignore or act in bad faith.

Trust is downstream of consequences

A society can sustainably ask people to trust third-party ecosystems around keys only if it can credibly demonstrate five things:

Investigations are independent. Abuses of power or discretion are be reviewed by entities not subordinate to the actor’s chain of command.
Consequences are real. If wrongdoing is found, discipline and prosecution must be plausible, not merely theoretical (or in recent years impossible).
Jurisdictional complexity is not a loophole. A system where accountability can be indefinitely delayed by forum fights teaches citizens to distrust outcomes (Day, 2026).
Digital coercion is constrained by due process. Seizure and compelled authentication should be rare, reviewable, and bounded—not casual and consequence-free.
The public can verify claims. Transparency matters because secrecy is where impunity breeds.

Pretti’s case is instructive because reporting has emphasized both the existence of video evidence and the competing narratives surrounding it, as well as the procedural and jurisdictional obstacles that can arise when state authorities confront federal actors (ABC News, 2026; Day, 2026; The Guardian, 2026). When citizens conclude—even mistakenly—that “even if I’m right, the system may not care,” they will extend that conclusion to digital life. They will assume devices can be seized. They will assume coercive unlocking is a practical possibility. They will assume third parties will comply first and explain later.

At that point, the society fractures. The well-resourced keep strong keys and strong legal defenses; the less-resourced are told to adopt “modern authentication” while living with the suspicion that convenience will be turned into pressure.

The civic bottom line

A society cannot sustainably tell citizens, “Trust us,” while also demonstrating—through visible, incidents followed by deceptive statements, as in the case of Alex Pretti's public execution—that coercive power may be applied without credible consequence. Without this trust of law and order, there can be no trust in the society which asks us to implicitly trust third party encryption.

So yes: ask whether we can trust security keys held by a third party under the current corrupt and lawless administration. But do not stop there. Ask whether we can trust the institutions that surround those keys—banks, email providers, vendors, etc., and the oversight structures which have business with the same administration and benefit from their cooperation or who may capitulate if pressured.

References

ABC News. (2026). A minute-by-minute timeline of the fatal shooting of Alex Pretti involving federal agents. https://abcnews.go.com/Politics/minute-minute-timeline-fatal-shooting-alex-pretti-federal/story?id=129547199

Apple. (2024, December 19). iCloud Keychain security overview. Apple Support. https://support.apple.com/guide/security/icloud-keychain-security-overview-sec1c89c6f3b/web

Blaze, M. (1994, August 20). Protocol failure in the Escrowed Encryption Standard. https://www.mattblaze.org/papers/eesproto.pdf

Cook, T. (2016, February 16). A message to our customers. Apple. https://www.apple.com/customer-letter/

Day, J. (2026). The unprecedented challenge of prosecuting federal agents for killing Renee Good, Alex Pretti. Minnesota Star Tribune. https://www.startribune.com/the-unprecedented-challenge-of-prosecuting-federal-agents-for-killing-renee-good-alex-pretti/601565078

FIDO Alliance. (n.d.). Passkeys. https://fidoalliance.org/passkeys/

Gibbs, N. (2016, March 18). Tim Cook: The Apple CEO speaks up. TIME. https://time.com/4258170/tim-cook-apple-ceo-interview/

National Institute of Standards and Technology. (1994, February 9). FIPS PUB 185: Escrowed Encryption Standard (EES). https://csrc.nist.gov/files/pubs/fips/185/final/docs/fips185.pdf

National Institute of Standards and Technology. (1994, June 6). Statement in response to Blaze key escrow paper. https://www.nist.gov/news-events/news/1994/06/statement-response-blaze-key-escrow-paper

Sforza, J. (2015, July 8). FBI director: Encryption challenge must be solved soon. Houston Chronicle. https://www.chron.com/neighborhood/fortbend/news/article/FBI-director-encryption-challenge-must-be-solved-9671365.php

Temoshok, D., Choong, Y.-Y., Regenscheid, A., Galluzzo, R., Fenton, J. L., Richer, J., & Lefkovitz, N. (2025). Digital Identity Guidelines: Authentication and authenticator management (NIST SP 800-63B-4). National Institute of Standards and Technology. https://doi.org/10.6028/NIST.SP.800-63B-4

The Guardian. (2026, January 24). Alex Pretti did not brandish gun, witnesses say in sworn testimony. https://www.theguardian.com/us-news/2026/jan/24/alex-pretti-killing-witness-testimony

United States v. Lavabit, LLC, 749 F.3d 276 (4th Cir. 2014). https://law.justia.com/cases/federal/appellate-courts/ca4/13-4625/13-4625-2014-04-16.html

U.S. Senate Committee on the Judiciary. (2015, July 8). Going dark: Encryption, technology, and the balance between public safety and privacy (S. Hrg. 114-905). Congress.gov. https://www.congress.gov/event/114th-congress/senate-event/LC74566/text

World Wide Web Consortium. (2021, April 8). Web Authentication: An API for accessing public key credentials (WebAuthn) Level 2 (W3C Recommendation). https://www.w3.org/TR/webauthn-2/

Zetter, K. (2013, October 3). Edward Snowden’s e-mail provider defied FBI demands to turn over crypto keys, documents show. WIRED. https://www.wired.com/2013/10/lavabit-unsealed/