Force TikTok divestment or ban for national security?

We can acknowledge real national-security concerns without handing the government a platform-sized off switch. China’s National Intelligence Law is not a bedtime story, and ByteDance’s 2022 scandal—employees accessing U.S. journalists’ data—was a red flag the size of a Jumbotron. But an outright ban is a blunt instrument that risks trampling the First Amendment and setting a precedent future officials could use on whatever app irks them next. Courts already swatted down earlier attempts to ban TikTok under IEEPA, and the 2024 divest-or-ban law will face strict scrutiny because it restricts where millions of Americans speak, organize, and do business. A smarter play is boring but effective: comprehensive data-privacy rules plus targeted national-security safeguards that apply to everyone. Pass robust, ADPPA-style privacy legislation that limits data collection and kills the data-broker free-for-all. Require algorithmic risk audits, transparency about state-linked content operations, and meaningful CFIUS oversight with enforceable mitigation—real data localization with independent trustees and penalties that bite. If there’s divestment, it must be technically credible (full code, data, and governance separation), not just a logo swap. Bottom line: defend national security with precision tools, not censorship by precedent. We can protect Americans from foreign influence without giving Washington the power to yank platforms like a bad DJ. If Congress wants to be tough, start by cutting off the data pipelines that make manipulation possible and apply those standards across the board—no special carve-outs for whichever app is trending on the Hill this week.

We can walk and chew gum: safeguard free speech and stop handing the CCP a direct line into American data and discourse. TikTok isn’t just another dancing app; it’s controlled by a company answerable to Beijing’s National Intelligence Law. We’ve already seen ByteDance employees misusing data, and security officials have warned about algorithmic manipulation—especially in an election year. The 2024 divest-or-ban law is precisely the kind of narrow, targeted remedy that says: keep the product Americans like, cut the foreign adversary control we don’t. Divestment is the market-friendly option. If ByteDance can decouple cleanly—source code, data, governance, and update pipelines—great. If Beijing blocks a sale, that tells you everything: it’s not about fun videos; it’s leverage. We already bar Huawei from our networks and Kaspersky from federal systems; treating a foreign adversary–controlled social platform as critical infrastructure isn’t radical, it’s overdue. “Project Texas” was a PR tour with holes—Oracle hosting doesn’t solve control of the algorithm or decision-making. Yes, pass a national privacy law; by all means starve the data brokers. But don’t let the perfect be the enemy of the urgent. A narrow standard for foreign adversary–controlled apps, transparent criteria, periodic review, and judicial oversight can protect speech while closing a glaring security gap. Free speech isn’t a suicide pact, and cat videos shouldn’t be a Trojan horse for a geopolitical rival.

New law, old problem: we’re still trying to fix a software supply chain and governance risk with a sledgehammer. The divest-or-ban statute effectively deputizes app stores and ISPs to preemptively pull speech platforms, then dares courts to sort out the constitutional rubble. That’s backwards. If the concern is covert influence or data exfiltration, regulate the behaviors—code provenance, data access pathways, and state-linked operations—not the passport of the parent company. Require continuous third-party code audits with tamper-evident logs, independent key management, and onshore incident response. Push ranking models toward on-device inference for sensitive signals, and mandate FARA-style labeling for state-affiliated content. Pair that with a real data-broker crackdown so no one can buy the same data TikTok could hypothetically steal. The bill’s designations are more elastic than advertised. Today it’s one app; tomorrow it’s a list that grows when geopolitics get spicy. That’s not a conspiracy theory—just an observation about how national-security authorities expand once created. And while we’re at it: nine to twelve months for a clean separation of code, training data, moderation policies, and update pipelines is fantasy scheduling. If Congress insists on divestment, build a credible playbook: full transfer of model weights and training corpus; re-staffed governance; documented dependency removal; and an 18–24 month runway with continuous verification, not a cliff that defaults to a ban. One more practical point: a ban-by-deadline hands a windfall to incumbent giants and tells every aspiring platform that Washington can yank distribution if your cap table spooks them. That’s not national security; that’s picking winners. We should not mirror the Great Firewall to own Beijing. Set a high, behavior-based standard for all high-reach platforms—domestic or foreign—aligned with the DSA-style risk audits our allies are already using, and keep the First Amendment as the guardrail, not a speed bump.

Round 2 reality check: this isn’t about punishing a passport; it’s about eliminating a control channel. You can encrypt logs, sign builds, and sprinkle auditors like confetti, but if corporate governance and update authority ultimately answer to a system that can compel cooperation in secret, you’ve engineered trust theater. The 2024 law cuts at the root—ownership and operational control—while preserving the product if ByteDance sells. That’s not overbroad; it’s narrowly keyed to foreign-adversary control defined by Congress, with judicial review and periodic reassessment. We’ve done this before: CFIUS forced Grindr’s sale over data risks, and no one confused that with book burning. On feasibility, the claim that divestment is impossible is convenient—but not credible. If Beijing says the algorithm is too strategic to export (as it hinted in 2020), that confession is Exhibit A for why we can’t leave it in place during U.S. elections. And if a sale proceeds, you don’t wing it: escrow the code and weights, rebuild the update pipeline in a U.S.-controlled clean room, rehire engineering leadership under domestic governance, and bar dependency on PRC-based contractors. That’s how you turn “logo swap” into a genuine fork. If ByteDance refuses or China blocks it, the fallback isn’t eternal dithering; it’s the off-ramp Congress wrote into law. Allies are moving, too. India banned the app outright years ago; the EU and Canada have already restricted it on government devices and opened investigations under the DSA. Pretending that a privacy bill alone will neutralize algorithmic manipulation ignores how influence operations actually work. By all means starve data brokers—and we should—but that’s orthogonal to a platform where a foreign adversary can tune the global microphone. Divestment keeps the competition and creators; it removes the CCP umbilical. That’s the balance: protect speech by protecting the system it runs on.

Call me old-fashioned, but when Congress writes a law that names a single platform class and threatens a nationwide speech blackout if a sale doesn’t happen on a political timetable, courts tend to reach for the First Amendment fire extinguisher. TikTok’s lawsuit in the D.C. Circuit isn’t a stunt; it’s a reminder that strict scrutiny asks whether there’s a less speech-restrictive way to achieve the goal. Montana’s statewide ban already flunked that test at the district court stage, and the federal statute’s “divest or disappear” mechanic is still a prior restraint in practical effect. If we actually want a durable fix that survives judicial review, we should regulate capabilities (data access, covert influence operations, and update control) rather than ownership provenance alone. Here’s the grown-up toolkit: a designated platform risk regulator (think DSA, but American) requiring continuous red-team audits of ranking systems, secured code provenance with hardware-backed attestations, independent researcher access with liability safe harbors, and mandatory incident reporting to CISA. Pair that with middleware choice—let users pick third-party ranking providers for the “For You” feed—to dilute any single operator’s influence. Add FARA-style labeling for state-linked content and algorithmic escrow that triggers supervisory control if a platform is credibly found to be manipulated by a foreign state. None of this is vibes; it’s verifiable, enforceable, and it applies across the market so we don’t play whack-a-passport while Temu, CapCut, or the next widget slips through. And about that rushed divestment: compressing a global codebase split, safety tooling transfer, and moderation retraining into nine months is how you ship regressions at scale. A botched fork could degrade trust-and-safety coverage right as we head into an election, which would be ironic if it weren’t dangerous. Geopolitically, a U.S. platform ban hands China a propaganda slide about the “American firewall” and encourages copycat crackdowns abroad. Protect national security, yes—but do it with rules that raise security and transparency for every large platform, not a one-off lever that normalizes government kill switches.

National security doesn’t wait for perfect paperwork. In the past year alone we’ve seen how fast narratives can be amplified during crises—the Israel–Hamas war, Ukraine, Taiwan tensions—and how recommendation engines can tilt the attention landscape. No one needs to prove a smoking-gun memo from Zhongnanhai to see the risk: a platform whose parent can be compelled in secret under China’s National Intelligence Law is a live control channel during an election year. Add to that the UK ICO’s sanction over kids’ data and the EU’s ongoing DSA probes: this isn’t a pristine actor being unfairly maligned. Divestment is the narrow remedy that removes the control channel while keeping the product. The implementation isn’t hand-wavy: escrow the code and model weights; rebuild the update pipeline in a U.S.-controlled clean room; migrate trust-and-safety tooling with continuity plans; seat a domestically accountable board with cleared security leads; and lock out PRC-based contractors. Put CISA and an independent technical monitor on continuous compliance with signed build attestations and penalties that actually sting. Courts will get their say—good. But the bill-of-attainder soundbite is weak when the statute defines a general, reviewable standard keyed to foreign adversary control, just like CFIUS-driven divestitures we’ve done before (see: Grindr). As for the “splinternet” lament, this is a defensive move, not an ideological Great Firewall. We already bar Huawei from our networks and Kaspersky from federal systems because control matters. Researcher access and middleware are fine as supplements; they are not substitutes for cutting the umbilical that lets a rival state quietly tune the microphone. If Beijing blocks a sale, it confirms the strategic value we’re worried about. If it greenlights one, we’ve preserved creator livelihoods and competition while closing a glaring security gap. That’s responsible sovereignty, not panic—and yes, you can still have your cat videos.

Even if you’re sold on divestment in theory, the statute’s execution is a constitutional faceplant waiting to happen. Packingham reminded us that social media is the modern public square; a nationwide cut-off tied to an executive designation tries to do platform governance with a national security wink and a prior restraint nudge. The litigation problem is practical, not just poetic: the government will lean on classified evidence that creators and users can’t meaningfully contest. Courts can peek in a SCIF, sure, but strict scrutiny still asks for narrow tailoring and less speech-restrictive means. For a law that makes distribution of a major speech forum contingent on a clockwork sale to pass that test, the government needs more than trust us in an election cycle. Now the operational own-goal: a ban does not erase demand; it reroutes it through gray markets. We will push millions toward sideloaded APKs, sketchy VPNs, and malware-adjacent clones that are actually harder to monitor and patch. Nothing says national security like driving teens and small businesses to Russian mirror sites and Telegram link trees. ISPs and app stores can yank an icon, but they cannot revoke network effects or curiosity. Meanwhile, forcing a cliff makes trust-and-safety coverage dip precisely when we need it steady, and collapses a venue heavily used by creators, diaspora communities, and small businesses for outreach. That is a security and resilience downgrade masquerading as firmness. If the goal is resilience against foreign manipulation, fix structure across the market instead of swinging a single-app wrecking ball. Mandate cross-platform interoperability and data portability so no one company is a single point of failure; require independent risk audits by an allied oversight consortium so the US, EU, and partners run common test suites on ranking engines; condition federal ad spend and procurement on meeting baseline security and transparency standards; and seat public-interest directors with security clearances on boards of systemically important platforms to oversee update pipelines and escalation during crises. Also, consider that a ban invites reciprocal measures under the digital trade rules we champion; the fastest way to legitimize copycat crackdowns abroad is to write the playbook at home. We can harden the whole ecosystem without normalizing government kill switches that future officials will be tempted to use.

This is a structural fix to a structural risk, not a gag order dressed up for court. Turner Broadcasting upheld content-neutral rules on gatekeepers when the government showed a real risk to the information pipeline; this law is the same genus, aimed at ownership and operational control rather than viewpoint. Judges can review a classified annex just as they do in CFIUS cases and FISA disputes. And the remedy keeps speech available if ByteDance sells; if Beijing blocks divestment, that is a bright-line signal about control that no middleware or audit can paper over. Holder v. Humanitarian Law Project also teaches that courts afford latitude where foreign influence and national security intersect, even when speech-adjacent conduct is regulated. On enforceability, let’s not pretend side-loading turns into a mass movement on iPhones or that gray-market APKs recreate the same reach. Real-world case study: after India’s 2020 ban, workarounds existed but the national-scale megaphone collapsed, and Reels and local alternatives filled the void. If we remove the official app from US app stores, cut CDNs and payment rails, and bar domestic ad networks and cloud contracts, you reduce the real-world impact from a stadium PA system to a guy with a megaphone in a parking lot. Friction matters in influence operations; it buys time and limits scale during crises. Concerned about incumbents getting a windfall? Good. Bake safeguards into the remedy: require creator data portability and export tools, mandate interoperability APIs for short-form video across major platforms for a transition period, and task DOJ and FTC to enforce conduct remedies on Meta and YouTube if market power spikes. Add a transition fund for small businesses that relied on TikTok ads, financed by civil penalties for noncompliance. To prevent authority creep, sunset the designation power with biennial reauthorization and independent technical audits. Finally, coordinate a G7 standard on foreign adversary platform control so we are not acting alone. Privacy reform is still worth doing, but it addresses ambient risk; divest-or-ban closes a specific, high-leverage control channel now. You can keep the cat videos; you just cannot keep the CCP umbilical attached to the microphone.

New angle, same caution: a divest-or-ban clock doesn’t just tug at the First Amendment—it yanks on supply chains and geopolitics. Beijing already telegraphed it views the TikTok algorithm as a strategic export; it restricted iPhones in state offices and squeezed Micron’s China market last year. Push a forced sale on a tight fuse and expect retaliation on U.S. firms from Cupertino to Boise, plus tit-for-tat rules that undercut the digital trade principles we lecture everyone else about. Yes, national security has a carve‑out in trade law—but if we wield it with a sledgehammer, we invite copycat “security” bans on American platforms abroad. If the real risk is covert control, build universal guardrails that don’t depend on flag-of-ownership and actually scale. Stand up a platform security regime—think FINRA-for-feeds—jointly run with CISA and NIST’s AI Safety Institute. Require reproducible builds, signed updates with hardware-backed attestations, SBOMs, tamper-evident log retention, and independent red-team exercises for ranking systems. Mandate crisis protocols: pre-declared change freezes around elections, independent holdback experiments on politically salient content, and rapid disclosure of state-linked ops with a FARA-style registry. Apply it to any high-reach platform—YouTube, Reels, X, TikTok, the next viral widget—so we don’t play whack-a-mole while influence hops to the next app. If Congress insists on leverage, use a conditional license with milestones instead of a speech cliff. Put a court-appointed technical monitor in the loop, stage control transfers (keys, update pipeline, data governance) across 18–24 months, and tie penalties to missed checkpoints: data minimization mandates, ad restrictions, and forced researcher access—not instant blackout. Add creator portability guarantees and an export tool for follower graphs so small businesses aren’t collateral. If Beijing blocks a credible separation, escalate proportionally—rate-limit new installs, require a follow-only default feed during campaign windows, restrict sensitive-data collection—before you swing the ban hammer. That sequence buys security without normalizing a government off switch future officials will find tempting for the wrong reasons.

Reciprocity isn’t a vibe; it’s policy. China’s National Intelligence Law can compel cooperation in secret, and its 2020 export controls hinted the TikTok algorithm is strategic enough to block a sale. That’s the control channel we’re closing. The House passed the divest-or-ban bill with broad bipartisan support because the risk is structural: a foreign adversary can quietly steer a platform that shapes U.S. information flow. Courts can and should review the record in camera, as they do in CFIUS and national‑security cases; this is a content‑neutral control remedy keyed to ownership, not viewpoint. Implementation doesn’t have to be a kamikaze update. Use a special master and a general license during the sale window: freeze high‑risk feature launches, force U.S.-only key signing and build servers, rotate credentials, and cut PRC-network access to production. Migrate trust-and-safety tooling and staff to a ring‑fenced U.S. entity under a domestically accountable board; escrow the code, model weights, and moderation guidelines; and keep a U.S.-based SRE crew on call for continuity. Run the divestiture through an antitrust filter—no sales to Meta or Google—and prioritize a buyer that preserves competition. Stand up a transition fund for small businesses and creators, financed by civil penalties for noncompliance, to keep livelihoods intact while the cap table changes. Retaliation risk is real—but it already exists, as Micron and U.S. app restrictions in China showed. That’s why we coordinate: align standards with the G7 and the EU’s DSA enforcement (which is already probing TikTok’s systems) so this isn’t a solo act. And yes, finish the sensitive-data outflow rules under the 2024 executive order and pass a federal privacy law so data brokers stop selling our lives by the gigabyte. But none of that substitutes for cutting a live control umbilical in an election year. The cleanest outcome is a sale that keeps creators and competition while severing Beijing’s leverage. If Beijing refuses, the fallback isn’t prayer—it’s a ban. We like dance challenges; we don’t like intelligence challenges.