One label, four variables: what Title II reclassification taught regulated buyers about classification risk
The FCC’s February 2015 vote reclassified broadband as a telecommunications service. That label set who regulated providers, what they owed, how rules were enforced and how long they lasted. Each moved again, as vendor labels inside regulated institutions can.
Looking Back revisits past weeks one at a time, up to the most recent, and asks what it meant for the banks, health systems and public agencies we serve. The series was written in 2026, with the benefit of hindsight, and each article says which week it looks back on.
Two-minute summaryThe dek and the first sentence of each section, quoted from the article
The FCC’s February 2015 vote reclassified broadband as a telecommunications service. That label set who regulated providers, what they owed, how rules were enforced and how long they lasted. Each moved again, as vendor labels inside regulated institutions can.
- The week On Thursday, February 26, 2015, the Federal Communications Commission voted 3 to 2 to adopt what it called the Open Internet Order.
- Jurisdiction: who oversees the provider Before February 2015, broadband was an information service.
- Obligations: what the provider owes The second variable is the content of the duties.
- Enforcement mode: bright lines and standards The third variable is how the duties are enforced.
- Durability: how long the answer lasts The fourth variable is time.
- The same four variables in the institution’s own vendor labels Regulated institutions classify their vendors every day, and each classification sets the same four variables.
- Objections, and where the argument is weakest One objection is that the analogy is too loose.
- With hindsight The 2015 classification lasted about three years.
The week
On Thursday, February 26, 2015, the Federal Communications Commission voted 3 to 2 to adopt what it called the Open Internet Order. Chairman Tom Wheeler and Commissioners Mignon Clyburn and Jessica Rosenworcel voted in favor. Commissioners Ajit Pai and Michael O’Rielly dissented. The formal title was a Report and Order on Remand, Declaratory Ruling, and Order, FCC 15-24, in GN Docket No. 14-28 (FCC release).
The vote ended more than a decade of argument about how to classify broadband. The Commission had twice tried to write open internet rules without treating broadband as a common carrier service, and courts had twice struck them down. The most recent defeat was the D.C. Circuit’s January 2014 decision in Verizon v. FCC. That court accepted that the FCC had some authority under Section 706 of the Telecommunications Act of 1996. It held that the agency could not impose common-carrier-style rules on a service it had classified as an information service (FCC release).
The February 2015 order took the route the court had left open. It reclassified broadband internet access service, the retail service people buy from cable, telephone and wireless companies, as a telecommunications service under Title II of the Communications Act. It treated mobile broadband as a commercial mobile service under Title III. It relied on Section 706 as an additional source of authority (FCC release).
On that foundation it set three bright-line rules: no blocking of lawful content, applications, services or non-harmful devices; no throttling on those bases; and no paid prioritization, including prioritization of a provider’s affiliates. It added a general standard against unreasonable interference with, or unreasonable disadvantage to, consumers’ and edge providers’ use of the internet, to be applied case by case. It expanded the existing transparency rule. Providers had to disclose promotional rates, fees, surcharges and data caps in a consistent format, plus packet loss and network management practices. Providers with 100,000 or fewer subscribers got a temporary exemption (FCC release).
The order also asserted, for the first time, authority to hear complaints about the terms on which mass-market broadband providers exchanged traffic with other networks. It did not apply the bright-line rules to those interconnection arrangements. The Commission said it would address disputes case by case under Sections 201 and 202 (FCC release)(80 FR 19738).
The Commission paired reclassification with forbearance, its statutory power to decline to enforce parts of the Act. The press release said the order would forbear from 27 provisions of Title II and more than 700 associated rules. Broadband providers would not face rate regulation, tariffs, last-mile unbundling or new Universal Service Fund contributions. The order kept core sections. They were 201 and 202 on just and reasonable practices, 208 and related sections on complaints and enforcement, and 222 on customer privacy. Also kept were 224 on pole attachments, 225 and 255 on disability access, and part of 254 on universal service (FCC release).
Section 222 needs a closer look, because it carries most of this article’s argument. The order kept the statute but forbore from applying the FCC’s existing customer proprietary network information rules to broadband. Those rules had been written for telephone calls. The order observed that they did not address the kinds of data a broadband provider holds, such as browsing history, and left a broadband-specific rule for later (80 FR 19738).
The record behind the vote was large. The FCC said nearly 4 million comments had been filed in the proceeding (FCC release). The full text was released on March 12, 2015, and published in the Federal Register on April 13, 2015. Most of it took effect on June 12, 2015, with some disclosure requirements waiting for approval under the Paperwork Reduction Act (80 FR 19738). The order defined broadband internet access service as a mass-market retail service. Enterprise services, virtual private networks, hosting and data storage services fell outside that definition (80 FR 19738).
Most commentary on the February 2015 vote was about traffic: fast lanes, throttling and video streaming. For a bank, a health system or a public agency, the traffic rules were the least of it. Banks, health systems and agencies mostly buy enterprise connectivity on negotiated contracts, which the order did not cover. Their customers, patients and residents use mass-market broadband, which it did cover, but the rules on blocking and throttling rarely touched their services directly.
What should have held a compliance officer’s attention was the mechanism. The FCC changed one label, the statutory classification of a service, and that change reset four separate variables at once. Each of those variables then moved again over the following decade. Regulated institutions meet the same mechanism in their own vendor relationships, where a single classification decides which rules a provider follows and which obligations fall back on the institution.
We treat the four variables in turn, then apply them to the institution’s own vendors, and close with objections:
- Jurisdiction: which regulator oversees the provider, and which one does not.
- Obligations: which duties attach, and which the regulator chooses not to enforce.
- Enforcement mode: whether the duties are bright lines or standards applied case by case.
- Durability: how long the classification and its consequences can be expected to last.
The four are distinct. A change in jurisdiction can leave duties unchanged in text but alter who enforces them. A duty can exist without a clear enforcement mode. And all three can be well designed and still short-lived.
Jurisdiction: who oversees the provider
Before February 2015, broadband was an information service. The Federal Trade Commission policed broadband providers’ privacy and data security practices under its general authority over unfair and deceptive practices. The FTC Act exempts common carriers subject to the Communications Act from that authority. Reclassification made broadband providers common carriers for their broadband service. The likely effect was that the FTC lost its authority over that service and the FCC gained it.
The exemption’s scope was itself contested. In 2016 a panel of the Ninth Circuit read it as status-based, meaning a company that was a common carrier for any service was outside FTC reach entirely. In February 2018 the full Ninth Circuit reversed that reading. It held the exemption activity-based, so a common carrier remained subject to the FTC for its non-common-carrier activities (Ninth Circuit). For the three years in between, the boundary between the two agencies’ authority was uncertain even for services that had nothing to do with broadband.
Why jurisdiction matters to a regulated institution
Jurisdiction decides who can investigate, what penalties apply, and what kind of rules exist. The FTC largely worked through case-by-case enforcement and consent orders. The FCC could write prescriptive rules and impose forfeitures. The same conduct by a provider could face very different consequences depending on which agency had it in view.
For institutions whose customers, patients and staff relied on mass-market broadband, the question was practical. A health system might want to know which agency it could turn to if a provider injected content into traffic to its patient portal. A bank might want to know who would act if a provider sold data revealing which customers visited its website. Before February 2015, the answer was the FTC. Between June 2015 and June 2018, it was the FCC for broadband service. After that, it was the FTC again.
Two regulators, two toolkits
The two agencies worked differently. The FTC’s privacy and security work was mostly enforcement after the fact. It brought cases against specific practices it judged unfair or deceptive, and settled most of them through consent orders that could run for many years. Its expectations emerged from the pattern of cases. The FCC, as a sector regulator, could write prescriptive rules ahead of time, require specific disclosures, and impose forfeitures for violations.
Neither approach is better in the abstract. Rules give advance notice and are easier to audit against. Case-by-case enforcement adapts to new practices without a rulemaking. For a provider, and for institutions watching providers, the difference is predictability. Under rules, the institution can read what is allowed. Under case-by-case enforcement, it learns what was not allowed when someone is charged.
What an institution can do about its public’s providers
An institution cannot choose the broadband providers its customers or residents use. It can prepare for the case where one of those providers does something that harms the institution’s service. Content injection into web pages, interference with a particular application, and the sale of data revealing visits to a sensitive site are all examples that were discussed publicly in this period.
Preparation has three parts. The first is detection: monitoring that can show whether a problem is concentrated among users of one provider or region, which a portal or app can usually infer from network addresses. The second is a record: a short incident file with dates, affected users and technical evidence. The third is a route: knowing which agency, at the time, takes complaints about that provider’s conduct. The route changed twice between 2015 and 2018, so the incident playbook needs an owner who checks it.
Most institutions will never file such a complaint. The value of the preparation is that, if a provider’s conduct does harm customers or patients, the institution can show it noticed, acted and escalated. Regulators of the institution itself tend to ask that question after the fact.
Jurisdiction gaps
Transitions create gaps. When one regulator’s authority ends before the other’s rules are in place, conduct can fall between them. The 2015 order created such a gap for privacy. The FTC could no longer act on broadband service, and the FCC had forborne from applying its telephone-era privacy rules while it wrote new ones (80 FR 19738). Section 222 still applied as a statute, and the FCC could enforce it directly. There was no detailed rule telling providers what it required for broadband.
The gap had practical edges. A provider deciding in mid-2015 whether to use customers’ browsing data for advertising had a statute that said carriers must protect customer proprietary information, an FCC that had said its old rules did not fit, and no new rule yet. A careful provider would have asked counsel and waited. A less careful one would have proceeded. Institutions whose customers were affected had no way to know which kind of provider they were dealing with, short of reading each provider’s privacy policy.
Regulated institutions run the same risk inside their own programs. When a function moves from one internal owner to another, or from one vendor to another, obligations can lapse in the handover. The FCC’s gap was visible and public. An institution’s gaps are often invisible until an examiner or an incident finds them.
Obligations: what the provider owes
The second variable is the content of the duties. Reclassification did not import all of Title II. Through forbearance, the FCC picked the provisions it wanted and set aside the rest: 27 provisions of Title II and more than 700 rules, according to its release (FCC release). What remained was a short list of core duties, chosen to support the open internet rules and a few consumer protections.
Forbearance is a powerful regulatory technique. It lets a regulator apply a broad statutory category while tailoring the obligations to the industry. It also means the obligations depend on the regulator’s continuing choices. Duties that are forborne today can, in principle, be applied later by a different Commission through a new proceeding.
The privacy duty without a rule
Section 222 required carriers to protect the confidentiality of customer proprietary information. For telephone service, the FCC had detailed rules about call records, consent and disclosure. For broadband, the order applied the statute but forbore from those rules, reasoning that they were built around voice calls and did not fit the data a broadband provider holds (80 FR 19738).
That left providers with a broad statutory duty and no detailed rule. Some may have read the gap as freedom. Others treated it as risk, because the FCC could enforce the statute through individual cases. For the institutions whose customers used those providers, there was no clear federal standard on what a provider could do with browsing data that revealed, for example, visits to a bank’s or a clinic’s website.
Forbearance has an internal equivalent
Institutions practice their own version of forbearance. A policy says all vendors handling customer data must complete a security assessment, and then a business unit gets an exception for a small vendor. A standard requires multifactor authentication, and a legacy application is exempted. Each exception is a decision not to enforce part of the institution’s own rules.
The FCC’s forbearance was at least public, reasoned and recorded in the order. Internal exceptions often are not. The practices that make forbearance defensible translate directly: a written rationale for each exception, a named approver, a scope, compensating controls where needed, and an expiry date that forces a new decision. Examiners and auditors routinely ask for the exception register, and an institution without one cannot show that its exceptions were deliberate.
Transparency as the most portable duty
Of all the obligations in the order, the transparency rule had the longest history. It dated from the FCC’s 2010 rules, and the court in Verizon v. FCC had left it standing (FCC release). The 2015 order kept it and expanded it, requiring disclosure of promotional rates, fees, surcharges, data caps, packet loss and network management practices (FCC release). Disclosure duties are less contested than conduct rules, because they tell customers what the provider does rather than restricting it.
For institutions that depend on the public’s connections, provider disclosures are useful data. A health system planning video visits can read the main local providers’ disclosures on data caps and network management, and design its service to stay within typical limits. An agency planning online services for rural residents can read the same disclosures for the providers in its area. The disclosures do not guarantee performance. They describe the conditions the institution’s service will meet.
What the order excluded
The definition of broadband internet access service was narrow by design. It covered mass-market retail service. Enterprise services, virtual private networks, hosting and data storage were outside it (80 FR 19738). So were some specialized data services that do not travel over the public internet, such as a dedicated heart-monitoring service, one of the examples the FCC gave (FCC release).
Those exclusions shaped what the order meant for regulated institutions. An enterprise circuit to a data center was outside the rules. A remote patient monitoring service delivered over a dedicated network might be outside them too. A clinician working from home on a consumer connection was inside them. Knowing which side of a definition a service falls on is the first step in knowing what any rule requires.
Obligations that attach to the buyer instead
Where a provider has no obligation, the buyer often does. A bank remains responsible under its own information security program for customer data it sends across any network. A HIPAA covered entity remains responsible for safeguarding electronic protected health information in transmission. An agency remains responsible under federal information security law for its systems’ security. When the regulator of the provider changes, those duties do not move. Encryption in transit, which these institutions largely adopted for other reasons, is the control that makes them independent of the provider’s classification.
Enforcement mode: bright lines and standards
The third variable is how the duties are enforced. The 2015 order mixed two modes. The three bright-line rules were clear prohibitions. The general conduct standard, against unreasonable interference or disadvantage, was deliberately open. The FCC said it would apply the standard case by case, guided by a list of factors (FCC release).
The order also set out how enforcement would work. The Commission would investigate formal and informal complaints. It would issue enforcement advisories and advisory opinions, and it created an ombudsman. Its Enforcement Bureau could seek technical opinions from outside bodies (FCC release).
Why the mode matters for compliance
A bright line is easy to comply with and easy to audit. A provider either blocks lawful content or does not. A standard is harder. A provider designing a new practice, such as a zero-rating plan that exempts certain services from data caps, cannot know in advance whether the FCC will judge it unreasonable. It can ask for an advisory opinion, wait for enforcement against someone else, or take the risk.
Compliance programs inside regulated institutions face the same choice whenever a rule is written as a standard. Reasonable security, appropriate safeguards and adequate documentation are all standards. They allow flexibility. They also shift the burden of interpretation onto the regulated party, and the meaning becomes clear only through enforcement actions and guidance over time.
Building a program around a standard
Institutions that live with standards tend to build the same tools. They keep a log of how they interpreted each standard and why. They track enforcement actions and guidance from their regulator and update the interpretation when the regulator’s view becomes clearer. They seek formal guidance where a decision is large and uncertain. They document the risk they accepted when they could not get certainty.
The FCC’s advisory opinion process offered providers one of those tools. For a regulated institution reading the order, it was a useful model. When a rule is written as a standard, ask the regulator for some route to a binding or semi-binding answer before committing.
What a standard costs the institution that relies on it
Standards also cost the people the rule is meant to protect. A bright line lets a health system know that a provider may not throttle its patients’ video visits. A general conduct standard tells it only that unreasonable interference is prohibited, and leaves the institution to argue, if it ever comes to that, that a specific practice was unreasonable. The institution must assemble evidence and wait for a regulator’s judgment.
That cost is the reason regulated institutions often prefer contract terms where they can get them. A contract can turn a standard into a bright line between two parties. A service-level agreement that specifies latency, packet loss and restoration time is a bright line, enforceable without a regulator. It cannot reach the public’s connections, but it can cover the institution’s own circuits and the services it buys for staff.
Complaints as the engine
The order relied heavily on complaints to bring problems to the FCC’s attention. That design puts weight on whoever notices harm first. Consumers rarely know why a service is slow. Edge providers, including institutions that run large online services, are better placed to see patterns across many users. A regime that runs on complaints rewards the parties who collect evidence and file them.
Interconnection as a pure standard
The order’s treatment of interconnection was the clearest example of enforcement by standard. The FCC asserted authority but wrote no rules. It said it would hear disputes under the just-and-reasonable requirements of Sections 201 and 202 and watch the market (80 FR 19738). For content providers and networks exchanging traffic with large broadband providers, the result was a right to complain with no advance statement of what conduct was acceptable.
Durability: how long the answer lasts
The fourth variable is time. The FCC described its rules as sustainable, grounded in multiple sources of legal authority (FCC release). The choice of words reflected the history. Two earlier attempts had been struck down, and the agency built this one to survive litigation.
It survived the first test. In June 2016, the D.C. Circuit upheld the order in United States Telecom Association v. FCC, a decision the en banc Ninth Circuit later noted in its own opinion (Ninth Circuit). Litigation was not the only risk, though. A rule that rests on an agency’s classification can be reversed by the same agency under a different majority, through a new rulemaking with a new record.
Classification as a policy choice
The Communications Act did not define broadband. The FCC had to decide whether it was more like a telephone service or more like an online information service. The press release called the reclassification decision fundamentally a factual one, and the order rested it on how customers understood broadband: as a transmission platform (FCC release)(80 FR 19738). The dissenting commissioners disagreed with that reading. When an agency has room to choose between two classifications, each with a defensible argument, the choice can change when the people making it change.
Under the legal doctrine of the time, courts deferred to an agency’s reasonable reading of an ambiguous statute. That doctrine let either classification survive review, which made the classification as durable as the agency’s majority and no more. Later events, covered in the hindsight section, show how that doctrine itself changed.
Staged effective dates
Durability also has a short-term dimension. The 2015 order did not take effect all at once. Most of it was effective on June 12, 2015. Some expanded disclosure requirements waited for approval under the Paperwork Reduction Act, and the small-provider exemption depended on a later bureau decision (80 FR 19738). A compliance calendar that recorded only the headline date would have been wrong for parts of the rule.
Regulated institutions see the same pattern in their own rules: phased compliance dates by institution size, provisions awaiting further guidance, and exemptions under review. The control is a calendar built from the rule’s text rather than from press coverage, with each provision’s date and any condition attached to it.
Signals that a classification may move
Some signals of instability were visible on the day of the vote. The order passed 3 to 2, and both dissenting commissioners issued statements opposing it (FCC release). A party-line vote on a contested legal question tells a planner that a change in the Commission’s membership could reverse the outcome. Petitions for review filed soon after adoption told the same planner that the courts would test it.
An institution can watch for the same signals in its own regulators’ actions. Some rules deserve a shorter planning horizon. Examples are a rule adopted over strong dissent, one that depends on a contested reading of a statute, one challenged in court soon after adoption, and one tied to a single leadership’s priorities. None of those signals means a rule will fall. Each means the institution should avoid building controls that only make sense if the rule stays.
Planning for a reversible rule
For a regulated institution, the practical lesson is to avoid building controls that depend on a rule likely to reverse. If a health system’s telehealth plan assumed that providers could not throttle video calls because of the 2015 rules, that assumption had a limited shelf life. If the plan assumed nothing about provider conduct and specified quality of service in its own contracts where it could, it did not depend on the FCC.
Remote work is the clearest current case. Many institutions now rely on staff working from home over consumer broadband, under whatever federal and state rules apply to those providers at the time. A remote-work standard that specifies minimum connection quality, a backup path such as a mobile hotspot for critical roles, and monitoring of session quality does not depend on any classification. It would have worked the same under the 2015 rules, after their repeal, and after the 2025 court decision.
Contracts are slower to change than regulations, but they change on the institution’s schedule. Where an institution cannot contract directly, as with residents’ and patients’ home connections, it can design services to degrade gracefully and monitor for problems by provider and region.
The same four variables in the institution’s own vendor labels
Regulated institutions classify their vendors every day, and each classification sets the same four variables. The difference is that the institution, not a federal agency, often makes the first call. A wrong label creates the same kinds of gap the FCC’s transitions did.
Health care: conduit or business associate
Under HIPAA, a vendor that creates, receives, maintains or transmits protected health information on behalf of a covered entity is generally a business associate. It must sign a business associate agreement and comply with parts of the HIPAA rules. HHS has long recognized a narrow conduit exception for services that only transmit information, with no more than transient storage, such as an internet service provider moving data. HHS guidance on cloud computing later made clear that a cloud provider storing protected health information is not a conduit, even if it cannot view the data.
That one label sets all four variables. It decides whether HHS can enforce against the vendor directly, which duties the vendor owes, whether the duties are specific or general, and whether the arrangement survives a change in the vendor’s service. A health system that labels a messaging or storage vendor a conduit has decided that the vendor owes it nothing under HIPAA. The decision should be documented, reviewed by counsel, and revisited whenever the vendor’s service changes.
Financial services: service provider classifications
Banks and credit unions must oversee service providers that access customer information under the security guidelines their regulators issued under the Gramm-Leach-Bliley Act. Non-bank financial institutions fall under the FTC’s Safeguards Rule. Broker-dealers have their own obligations under SEC rules. Whether a vendor counts as a service provider, a critical vendor or a material outsourcing arrangement decides how much diligence and monitoring the institution owes.
Those classifications also move. Regulators revise guidance on third-party risk, and a vendor’s role can expand from a narrow service to one touching customer data. An institution that set a vendor’s tier at onboarding and never revisited it has the same durability problem as a rule built on an agency classification.
Government: what kind of information, on what kind of service
Agencies classify both the information and the service. Criminal justice information brings the FBI’s CJIS Security Policy into any system that handles it, whoever runs that system. Federal agencies using cloud services generally require FedRAMP authorization at a level matched to the data. A service classified as handling low-impact data, then expanded to handle moderate-impact data, needs a new decision. The label sets which rules apply, who enforces them, and how specific they are.
New services, new labels
The same question recurs with every new kind of service. When institutions adopted cloud storage, they had to decide whether the provider was a business associate, a service provider handling customer information, or a system requiring FedRAMP authorization. When they adopted collaboration tools, they had to decide whether messages were records and whether the vendor was a recordkeeping system. Since 2023, generative AI services have raised the same decisions again, with an added one: whether the vendor may use the institution’s data to train its models.
In each case, the first classification tends to be made quickly by whoever buys the service. It sets the four variables for that vendor. If it is wrong, the institution may have no agreement where one was required, the wrong diligence, and no plan for when the vendor’s service changes. Making the classification a deliberate decision, recorded with reasons, is the cheapest point to get it right.
A register of classification decisions
The practical tool is a register. For each significant vendor, it records the classification under each relevant regime, the reasoning, the date, the owner, and the triggers that would require a new decision. A change in the vendor’s services, a change in the data it touches, a change in the regulator’s guidance or a change in law all qualify. The register makes the four variables visible for every vendor, which is what the FCC’s public record did for broadband.
Running the register is modest work. The fields are few: vendor, service, data touched, classification under each regime, reasoning, decision owner, date, review triggers and next review date. The review cadence can follow the vendor’s risk tier, annual for the most critical and longer for the rest. The useful metrics are the share of significant vendors with a current classification and the number of reviews triggered by service changes, which shows whether the triggers work.
Ownership is usually split. Procurement knows the vendor, the privacy or compliance office knows the regimes, and the business owner knows how the service is used. The register works when one of them, usually compliance, owns the record and the others are required to report changes.
The register also helps when a regulator changes its own classification. An institution that knows which vendors it classified under which reading of a rule can find the ones affected within days. Without it, the search starts from contracts and memory.
Differences within each sector
The variables land differently by institution type. A national bank and a state-chartered credit union follow similar security guidelines through different regulators, the OCC for one and the NCUA and the state for the other. A broker-dealer has its own recordkeeping and privacy rules under the SEC and FINRA, which classify communications and vendors in ways a bank’s rules do not. A health system with its own health plan has both covered-entity roles to manage, while an ambulatory network may have only one. A federal agency follows federal information security law and FedRAMP. A county may follow state law and, for some systems, federal program rules tied to grants. Each regime has its own labels, and an institution under several regimes holds several classifications for the same vendor.
That multiplicity is why a single vendor tier often hides the full picture. A vendor can be low-risk for one regime and a business associate under another. The register should record each classification separately rather than collapse them into one score.
Objections, and where the argument is weakest
One objection is that the analogy is too loose. The FCC classifies an entire industry by rulemaking. An institution classifies individual vendors by contract and internal decision. The scale and process differ. We accept the difference and think the analogy still holds where it matters. In both cases, one label sets jurisdiction, obligations, enforcement mode and durability together, and in both cases the label can be wrong or can change.
A second objection is that the 2015 order did not matter to regulated institutions at all, because their own circuits were enterprise services outside it. That is largely true for the institution’s own connectivity. It is less true for the public side: customers, patients and residents on mass-market connections, and staff working from home. And it is not true for the mechanism, which recurs in every regulatory regime these institutions operate under.
A third objection is that classification instability is a feature of communications law and rare elsewhere. The record since 2015 suggests otherwise. Doctrines on agency deference have shifted, and the treatment of cloud services, data brokers and third-party providers has changed under several regimes. We would not claim every classification is unstable. We would claim that any classification resting on an agency’s interpretation of an ambiguous statute should be treated as revisable.
A fourth objection is cost. A classification register, provider monitoring and contract work all take staff time, and smaller institutions have little to spare. We would scale the work to the risk. A credit union with a handful of critical vendors can keep the register in a spreadsheet maintained by its compliance officer. A large health system will want it inside its third-party risk management system. The principle is the same at either size: the classification is a decision someone owns.
The argument is weakest where an institution has little choice. A rural clinic with one broadband provider cannot contract around provider conduct, whatever the rules say. For that clinic, the federal classification matters more than any of the controls described here, and its realistic response is monitoring and contingency planning rather than contract terms.
With hindsight
The 2015 classification lasted about three years. The D.C. Circuit upheld the order in June 2016. In the meantime the FCC used its new authority. In March 2016, its Enforcement Bureau settled an investigation of Verizon Wireless’s insertion of unique identifier headers into customers’ mobile traffic, an inquiry that cited Section 222 and the 2010 transparency rule. Under the consent decree, Verizon Wireless paid $1.35 million and agreed to get customers’ opt-in consent before sharing the identifiers with third parties for targeted advertising (FCC, DA 16-242).
In October 2016, the FCC adopted the broadband privacy rule the 2015 order had deferred. It was published on December 2, 2016, as “Protecting the Privacy of Customers of Broadband and Other Telecommunications Services.” It never took effect in full. Congress disapproved it under the Congressional Review Act, and the President signed the resolution on April 3, 2017, as Public Law 115-22. The law provides that the rule shall have no force or effect (Pub. L. 115-22). Under the Congressional Review Act, an agency generally may not reissue a disapproved rule in substantially the same form without new legislation.
In December 2017, a new FCC majority reversed the classification. Broadband became an information service again, and mobile broadband a private mobile service. The conduct rules were removed, and the change took effect on June 11, 2018. Jurisdiction over providers’ privacy practices returned to the FTC. Earlier that year, in February 2018, the en banc Ninth Circuit had held the FTC Act’s common carrier exemption to be activity-based (Ninth Circuit). That ruling meant the FTC’s reach did not depend entirely on a company’s status as a carrier.
In April 2024, the FCC reclassified broadband as a telecommunications service and reimposed rules similar to those of 2015. Two months later, in Loper Bright Enterprises v. Raimondo, the Supreme Court overruled the doctrine of deference to agencies’ reasonable readings of ambiguous statutes. On January 2, 2025, in Ohio Telecom Association v. FCC, the Sixth Circuit set aside the 2024 order. It held that broadband is an information service and mobile broadband a private mobile service under the Act, reading the statute itself rather than deferring to the agency (CRS).
The Congressional Research Service noted that several states had adopted their own rules by legislation or executive order, naming California, Washington and New York (CRS). At the time of writing, in November 2026, no federal open internet rule of the 2015 kind is in force, and the main federal check on broadband providers’ data practices is the FTC’s general authority.
What held up? The four-variable reading did. Each variable moved at least once. Jurisdiction passed from the FTC to the FCC and back. The privacy obligations went from statute without a rule, to a rule, to no rule. Enforcement shifted from bright lines to no conduct rules. Durability was shorter than the FCC’s own word for its rules suggested.
What changed was the reason classifications move. In 2015, the risk was a change of agency majority, because courts deferred to either reading. Since Loper Bright, courts decide the best reading of the statute themselves. That makes a classification more stable once a court has ruled on it, and less dependent on who leads the agency. It also means agencies have less room to adapt old statutes to new services, which shifts more of the work to Congress and, for regulated institutions, to their own contracts.
Spotted an error? Tell us and we will check it and correct the article where needed.