Adak Island Internet Subsidy Exposes FCC Audit Gaps

The hard part of the Adak Island internet subsidy story is not believing that broadband is expensive in the Aleutians. Of course it is. The hard part is explaining why a federal support program kept paying more than $340,000 a year for internet service tied to 306 locations on an island where many of those locations were vacant or deteriorating, where no resident interviewed by reporters subscribed to the subsidized service, and where residents instead described using Starlink for everyday connectivity.[1]

That is the useful starting point for the government contract and subsidy questions now facing the Federal Communications Commission. The visible waste is easy to understand: public money assigned to buildings that, in many cases, no longer appear to serve the public. The harder question is administrative. Which control was supposed to notice that subsidized availability had drifted away from actual use, and why did that control not change the payment?

The ProPublica and Anchorage Daily News investigation published on July 20, 2026, is recent enough that any legal assessment has to stay provisional. It is not a charging document, and it does not by itself prove fraud. But it does identify a fact pattern that compliance officers and counsel should recognize immediately: a recipient may remain inside the architecture of a fixed-support program while the public-purpose rationale for continued payment becomes increasingly difficult to defend.

The Subsidy Was Paying for Availability That Residents Were Not Using

Adak is not a normal service territory. It is a remote former Navy base in the Aleutian chain, with infrastructure built for a much larger Cold War population than the island now supports. Population estimates differ materially: Alaska Department of Labor figures cited in the investigation put Adak at 77 residents in 2025, while residents interviewed by reporters described the year-round population as closer to roughly two dozen.[1]

That distinction matters because federal support programs often work with location counts, buildout obligations, and service availability metrics rather than household-by-household demand. In Adak, the investigation reported that Adak Eagle Enterprises received more than $340,000 annually in federal broadband support to provide service at 306 locations, many of them vacant, damaged, or collapsing buildings left from the military era.[1]

The location count had already been adjusted once. The FCC previously required service to 346 locations, then reduced the number to 306 in 2022 after the company reported that some buildings were uninhabitable. The subsidy amount did not go down with the reduced obligation.[1]

That is the program-design failure in miniature. If a carrier’s required universe of subsidized locations shrinks because some locations are not habitable, a payment system concerned with current public benefit would at least ask whether the amount of support should shrink too. A fixed-support structure can answer that question differently. It can preserve the negotiated payment even after the performance target changes.

The demand signal was even worse. Reporters said they did not find a single interviewed resident who subscribed to Adak Eagle’s subsidized internet service. Those residents used Starlink instead. The investigation also reported the speed contrast as 1 to 10 Mbps for the subsidized service versus 447 Mbps for Starlink in a local test, and noted that the roof of Adak Eagle’s own office reportedly had two Starlink dishes.[1]

There is a temptation to treat that last detail as the whole story because it is so vivid. It should not be. The stronger point is not that a satellite dish on an office roof looks embarrassing. The stronger point is that subscriber behavior, building condition, and service performance all appear to have been available as warning signs, while the payment stream continued.

The Warning Signs Were Not New

The Adak record was not a clean file until Starlink arrived. In 2013, the FCC found Adak Eagle’s spending unreasonable, including an unnecessary vehicle fleet and a fishing boat, and found a $237,455 salary unreasonable and disproportionate, according to the ProPublica and Anchorage Daily News account of the order.[1]

That kind of finding should have changed the risk profile of the recipient. It does not automatically mean every later dollar was improper. But for a subsidy administrator, a prior reasonableness finding is not trivia. It is a marker that later payment integrity reviews should be more skeptical, not less.

The investigation also described a 2018 FCC Office of Inspector General audit. The available reporting does not establish that the audit made findings that required termination of support, and it would be a mistake to write as if it did. The narrower and better-supported point is that the carrier had already been visible to federal oversight actors before the 2022 location-count reduction and before the 2026 reporting on nonuse.[1]

Each signal has limits. Reporters’ interviews are not a census. Population estimates vary. A speed test is not a full engineering study. The subsidy totals reported for Adak Eagle, including $3.6 million in federal broadband support since 2016, $3.54 million in intercarrier compensation, and $3.75 million from an Alaska state fund, come from the investigation and should be treated as reported figures rather than independently reconstructed payment ledgers.[1]

But oversight does not require courtroom-level certainty before it asks basic program questions. If the official payment formula keeps generating the same payment after the location obligation is reduced, after the relevant buildings are described as uninhabitable, and after the supposed beneficiaries are using a different service, the problem is no longer just a recipient’s billing posture. It is the government’s measurement problem.

Fixed Support Can Create Compliance Comfort Without Creating Accountability

The Alaska Plan is important here because it helps explain how a payment can survive facts that look absurd from the outside. The program used a fixed 10-year support structure in which carriers negotiated performance obligations, and the reporting cited former FCC deputy chief Carol Mattey describing the structure as a result of political pressure against applying standard cost models in Alaska.[2]

There are real reasons to be cautious with standard models in Alaska. Remote villages, island communities, weather, transport costs, and thin customer bases can defeat mainland assumptions. A service obligation that looks uneconomic in an urban cost model may be exactly the obligation Congress had in mind when it built universal service policy.

But rejecting a cost model is not the same as rejecting measurement. A program can accommodate high-cost geography and still ask whether supported locations are habitable, whether residents are subscribing, whether speeds remain meaningful, whether cheaper or better alternatives have changed the public need, and whether a changed obligation should change the support amount.

This is where “we followed the rules” becomes an incomplete answer for regulated entities. In a fixed-support program, the rules may permit payment even when usage is negligible. That may reduce one category of legal exposure. It does not eliminate audit risk, reputational risk, legislative risk, whistleblower risk, or the possibility that regulators later reinterpret what certifications and performance reports implied.

For counsel, the distinction is practical. A carrier might be able to say, accurately, that it accepted support under an approved plan and met the plan’s formal reporting conditions. That does not answer whether internal emails, board materials, subscriber data, or field reports showed that the supported service no longer delivered the public benefit that agency reviewers assumed it delivered.

Adak Fits a Broader Alaska Oversight Pattern, But It Should Not Be Overread

The broader Alaska record gives the Adak facts more weight without turning every Alaska subsidy into the same case. Anchorage Daily News reported in May 2026 that the FCC had sent $4.6 billion to Alaska telecom companies since 2016 under programs supporting rural and high-cost service.[2]

Large spending is not proof of waste. Alaska is exactly the kind of place where large support numbers may be necessary. The relevant question is whether the administrator can tell the difference between high-cost service that is working and high-cost service that has become nominal.

The May investigation also reported that Summit Telephone received more than $1 million a year while owner Roger Shoffstall served federal prison time for tax evasion and ran the company from jail, and that the FCC did not cut off the subsidies.[2]

Again, the legal point is not that imprisonment for a tax offense automatically proves telecom subsidy noncompliance. It is that payment systems capable of continuing unchanged through facts that would alarm any ordinary grant officer need unusually strong monitoring logic. If the logic is simply that the support amount was set at the beginning and remains payable absent a narrow disqualifying event, the program has chosen administrative stability over responsiveness.

GCI is a different pressure point. In 2023, GCI agreed to pay $40.2 million to resolve False Claims Act allegations involving inflated pricing, bid rigging, and revenue underreporting in the FCC’s Rural Health Care Program; a former director who brought the whistleblower case received $6.4 million.[3] ADN later reported that GCI received $466 million in support two years after that settlement.[2]

That does not make the later support unlawful. Settlement of one program’s allegations does not necessarily bar a carrier from all future support. But for compliance professionals, the sequence is a reminder that federal program participation often turns on specific eligibility and performance rules, not on a general moral fitness test. That is why the rules need to measure the right things.

The Enforcement Question Is Sharper Because the FCC Has Used Stronger Tools Elsewhere

The Adak facts become more pointed when compared with the Sandwich Isles matter in Hawaii. There, the FCC fined Sandwich Isles Communications $49.6 million and ordered $27 million repaid, according to the Alaska reporting’s comparison of Universal Service Fund enforcement precedents.[2]

The comparison should be used carefully. Enforcement cases turn on different records, different rules, different certifications, and different evidence of knowledge. But the precedent matters because it shows the FCC is not categorically powerless when it concludes that universal service support has been abused.

So the question is not whether Adak should automatically produce a Sandwich Isles-style order. The question is why the Adak pattern did not produce, at minimum, a more visible subsidy adjustment, usage review, performance-trigger inquiry, or public explanation. Agencies do not need to punish every bad-looking file. They do need to explain why a file with multiple red flags stays on autopilot.

False Claims Act Risk Depends on More Than Waste

The False Claims Act is the wrong tool for expressing general disgust. It is also a mistake to assume it has no role in federal telecom subsidy programs. The GCI settlement shows that FCC-administered support programs can generate FCA exposure where allegations involve false pricing, bid manipulation, underreported revenue, or other misrepresentations tied to payment.[3]

Adak, as reported, presents a harder FCA theory. If the Alaska Plan fixed the support amount in advance and did not require active subscriber use at each supported location, then “the government paid for service residents did not use” may describe waste without establishing falsity. The legal analysis would need to examine the carrier’s certifications, performance reports, location submissions, knowledge of building condition, communications with the FCC, and any statements about service availability or capability.

For government contractors and subsidy recipients, that boundary is not comforting enough to ignore. Many FCA cases begin where a formal claim for payment intersects with internal knowledge that the operational reality is different from what the agency has been led to believe. A fixed-support structure may narrow the path to liability, but it does not make internal evidence irrelevant.

Counsel reviewing a comparable file would want to know several things before reaching a legal conclusion:

• What exactly did the recipient certify each year about locations, service availability, performance, and continuing eligibility?
• Did the recipient know that listed locations were vacant, collapsing, inaccessible, or otherwise unlikely to serve residents?
• Did any filing or communication imply active public benefit beyond what the fixed-support rules actually required?
• Did agency staff have the same information, and if so, did they expressly approve continued payment?
• Were there internal objections, whistleblower complaints, audit requests, or suppressed subscriber data?

Those questions matter because they separate bad program design from actionable misrepresentation. Both can waste public money. Only some fact patterns become enforcement cases.

The Audit Gap Is About What the Program Chose Not to Measure

Daniel Lyons of Boston College Law has testified that “the FCC is not very good at auditing its program,” and has advocated direct-to-consumer subsidy models, according to the Anchorage Daily News account.[2] The Adak case gives that criticism a concrete form. The problem is not simply that auditors failed to find one bad invoice. It is that the program appears to have lacked a routine mechanism for connecting payment to current usefulness.

Availability metrics are attractive because they are administrable. A carrier can report that it serves a location. An agency can compare that location to a map. A payment system can operate without asking whether a particular household subscribed last month. That simplicity has value in remote areas where demand can fluctuate and service capacity needs to exist before customers can rely on it.

But the Adak facts show the limit of that approach. When the listed locations are mostly remnants of a depopulated military installation, “available at a location” can become a formal answer to the wrong question. The public purpose is not to keep a spreadsheet green. It is to make communications service realistically available to people and institutions that need it.

A more responsive audit would not have to punish carriers for every low-take-rate area. It could use triggers: a material location-count reduction, evidence that supported buildings are uninhabitable, persistent zero or near-zero subscriber uptake, large speed gaps between subsidized service and widely adopted alternatives, or prior agency findings that the recipient’s costs were unreasonable. Any one trigger might justify inquiry. Several together should be difficult to ignore.

What Counsel Should Take From Adak

For legal departments, the lesson is not limited to telecom. Fixed subsidies, negotiated performance targets, cost-reimbursement arrangements, grant milestones, and regulated-rate mechanisms can all create the same false sense of finality. The agency approved the formula, so the recipient treats later payment as routine. Meanwhile, operational facts drift away from the premise that justified the formula.

The useful compliance move is to audit against both the rule and the purpose. That does not mean inventing obligations the contract or program order does not contain. It means identifying where the file would look indefensible if an inspector general, journalist, relator, appropriations staffer, or agency enforcement lawyer read it cold.

A subsidy recipient in a comparable position should be able to answer, in writing, why continued payment remains reasonable after material changes in demand, location condition, alternative service availability, or performance. If the only answer is that the payment formula has not yet been amended, counsel should treat that as a governance issue even if it is not yet a litigation issue.

For agencies, the Adak file is a warning about audit design. A program can be constitutional, authorized, and well-intentioned while still measuring the wrong proxy. The Supreme Court’s 2025 decision in FCC v. Consumers’ Research left the Universal Service Fund’s funding mechanism standing, but that constitutional survival does not resolve whether the FCC’s governance, auditing, and performance controls are adequate for the money being collected and spent.

The Successor Program Cannot Just Rename the Same Controls

The next practical forum is the Alaska Connect Fund rulemaking, the successor structure now moving after the Alaska Plan. That is where the Adak lesson should be translated into payment design rather than outrage.

A successor program can still recognize that Alaska is expensive and that high-cost communities need support before private demand alone can sustain service. But if it continues to pay for nominal locations without usage-sensitive auditing, enforceable performance triggers, and meaningful subsidy adjustment when obligations shrink or public benefit disappears, it risks reproducing Adak under a new program name.

References

1. You Pay for Internet Service in Empty Buildings on Alaska's Adak Island — ProPublica / Anchorage Daily News, July 20, 2026.
2. The FCC sends billions to Alaska companies selling slow internet. You pay for it. — Anchorage Daily News, May 19, 2026.
3. The FCC Won't Let Me Be — Arnold & Porter, May 2023.