Week ending July 17, 2026

The Commission spent this week on the two transitions that actually reshape the voice network — the move off legacy TDM interconnection, and the tightening of who gets to be a trusted participant in what replaces it. The Wireline Competition Bureau ran a two-day workshop on the all-IP transition, the interconnection docket drew ex partes from AT&T, Lumen, and the rural carriers, and the Robocall Mitigation Database FNPRM landed on the July 22 meeting agenda. They are the same story told at two layers: how calls get carried, and how the carriers get trusted.

At the IP-transition workshop: the move to a unified telephone network over IP

I spent Tuesday afternoon on the FCC’s “Consumer Protection and the IP Transition” panel, one of eight at the Wireline Competition Bureau’s two-day IP Transition Workshop (July 15–16, in the Commission Meeting Room). I was there for Somos, alongside panelists from Telecommunications Access of Maryland, Numeracle, Granite, and TNS. The workshop worked through the reforms now pending across the linked docket cluster — Advancing IP Interconnection (WC 25-304), Reforming Legacy Rules for an All-IP Future (WC 25-311), and Accelerating Network Modernization (WC 25-208). The real question on the table isn’t whether the transition happens but whether the FCC should push the industry to finish it on the Commission’s proposed timeline — roughly a two-year runway, with forbearance from the legacy TDM rules teed up for the start of 2029 — and what has to be true for the consumer protections wired into the legacy regime (911, accessibility, service continuity) to survive that pace.

Across the two days (agenda), the same question surfaced at four different layers. The interconnection panels worked the mechanics and the economics — the framework for handing off an all-IP call, and the hard tail of the transition, since much of the network has already migrated (on the order of 75–80%) and what remains is the routes where IP interconnection isn’t yet economical. The public-safety panel took up the NG911 transition and the gap between the 911 authorities ready to move and those still stranded on legacy infrastructure. And the consumer-protection panel took up accessibility and service continuity. The throughline Chairman Carr set in his opening was that getting the transition right strengthens the infrastructure and improves resilience — but the sharper version, which kept re-emerging, is that the all-IP network isn’t merely a faster voice network; it’s the substrate that finally makes verifiable identity, rich call data, interoperable NG911, and accessible communication native rather than bolted on.

My own panel is where the authentication argument lives, and it’s the point I spent the most time on. The discussion was grounded in where call authentication actually stands — and the honest answer is uneven: STIR/SHAKEN signing is near-universal between the Tier-1 carriers but drops off sharply below them, full A-level attestation remains the exception rather than the rule, and even a fully signed call loses its authentication at the seam bad actors route through — TDM in the middle, a legacy hop between two IP endpoints that strips the signed identity and the rich call data along with it. My argument was that the IP transition and call authentication are not two initiatives competing for attention but one architecture: bringing them together, with certificates as the trust layer, is what makes the association between a telephone number and the entity authorized to use it cryptographic and end-to-end rather than inferred hop by hop. That has direct consequences the record should hold onto. The identity, the attestation, and the branded call data a consumer actually sees survive end to end only if the path is end to end — every TDM segment is a place that information drops. B- and C-level attestation were never meant to be permanent; they were accommodations for an incomplete network, and finishing the transition is what moves the call population toward genuine A-level attestation. That connects straight to the Commission’s know-your-upstream-provider work, which pushes accountability down to the providers actually originating and signing traffic — enforceable precisely because an unbroken IP path is what lets the signature travel with the call. I pressed the numbering side of the same point: as the network stops being geographic, the static assumptions built into numbering — a number tethered to a rate center, an identity inferred from where a call appears to originate — stop paying their way, and an all-IP fabric lets numbering become dynamic with the number-to-entity binding asserted cryptographically rather than presumed from legacy plumbing.

The accessibility half of the panel was the part I care about most, and it was a genuine pleasure to sit alongside David Bahar of Telecommunications Access of Maryland, with whom I co-chaired the NANC Interoperable Video Calling work years ago — my own first real experience of video relay was in conversations with him. The same architecture that carries identity end to end carries accessibility end to end: an all-IP network can make real-time text, video relay, and NG911 native capabilities of a more unified telephone network, delivered through the telephone number itself rather than grafted onto a voice path never designed for them — and the transition is the moment to design for those users deliberately instead of retrofitting later. It’s the least-covered corner of the transition in the filings and the one that matters most to real users: TTY-to-RTT migration, relay-service continuity, and disability-access obligations carrying forward rather than lapsing when the underlying service retires.

The interconnection record: what replaces the legacy rules

Away from the workshop, the written record moved the same week, and it clarified the real fault line: not whether to finish the transition, but what replaces the legacy interconnection rules. Lumen, meeting July 13 with Chairman Carr’s senior counsel, pressed for the Commission to forbear from the legacy Section 251(c) TDM interconnection and collocation regime as of January 1, 2029 and to “stand up a new, light touch regulatory framework for IP interconnection for voice traffic.” It tied that to intercarrier-compensation reform, arguing that completing the move to bill-and-keep “would support the IP transition by eliminating legacy TDM compensation rules that distort incentives and encourage providers to maintain outdated TDM arrangements in order to preserve revenue flows.” The Lumen/USTelecom plan would require every voice provider to establish two points for receiving traffic — one reachable over the public Internet without an individualized contract, and one available through a negotiated arrangement.

AT&T filed on July 14 to defend that market-oriented model against the main competing proposal — Inteliquent’s National Transit Provider framework, which former Commissioner Furchtgott-Roth (writing for Hudson) had cast as the light-touch default. AT&T argued the opposite: the “Agreement-less IP Interconnection using the Internet model” needs “no new tariffing, no new rate design, no new intercarrier compensation regime,” while the NTP approach “would effectively graft legacy Section 251 and 252 concepts onto IP networks,” reviving “potential opportunities for regulatory arbitrage.” AT&T’s technical point is the one worth holding onto: responding to Professor Scott Jordan’s critique, it reported that over-the-Internet interconnection on its AT&T Phone – Advanced service ran a median latency of 41 ms and 0% median packet loss on unprioritized, best-efforts traffic — its evidence that the open-Internet path is “a real, scalable, ‘good quality’ solution” and not a downgrade. Notably, AT&T flagged that the NTP tariffed-transit model would also force the Commission to “evaluate how such a framework would affect providers’ ability to protect their customers from unwanted and illegal robocalls” — a direct acknowledgment that the interconnection architecture and the robocall-authentication stack are not separable problems.

WTA – Advocates for Rural Broadband, meeting July 15 with Commissioner Trusty’s advisor, supplied the through-line that ought to frame the whole beat. WTA argued that the interconnection, intercarrier-comp, 911, robocall, and universal-service proceedings “are interrelated” and “should all be addressed in a holistic fashion to ensure there is a smooth transition to all-IP networks without any adverse effects on rural customers,” anchored in the Section 254(b)(3) mandate that rural rates and services stay “reasonably comparable” to urban ones. It named the parallel tracks explicitly — NG-911, robocalling, USF support, telephone access charges — as facets of a single transition, and warned that a rural carrier “will incur significant costs, some of which will be ongoing, before it realizes any savings” from going all-IP. That holistic frame is the correct one: the transition is not an interconnection story with a 911 footnote; it’s one transition whose identity, safety, and access consequences all move together.

The RMD FNPRM heads to a July 22 vote

The trust layer moved in parallel. The draft Improving the Effectiveness of the Robocall Mitigation Database FNPRM — FCC-CIRC2607-04, released July 1, cross-captioned WC 24-213 and 17-97 and CG 17-59 — went onto the July 22 open-meeting agenda. It builds on last year’s RMD Report and Order by clarifying which entities must file, demanding more specific mitigation detail, trimming exemptions, and seeking comment on keeping removed bad-actor providers and their principals from re-entering the database under a fresh corporate shell. It also proposes to align RMD screening with the Commission’s know-your-customer obligations — the tell that the KYC ambitions everyone expected to route through the RMD are, in fact, doing so.

Numeracle filed the most substantive ex parte of the RMD cycle, meeting cross-bureau staff July 14. Its framing is blunt: “The Robocall Mitigation Database as currently implemented is not effective,” and “right now, the RMD contains dubious data leading to bad decisions.” The problem is structural — “RMD filings and robocall mitigation plans are self-assertions with no independent verification,” “over 11,000 filings must be individually reviewed by downstream carriers,” and “currently, a ‘D’ is a passing grade.” What elevates the filing is the gatekeeper argument: it lays out six separate credentialing chokepoints a provider already clears to originate traffic — FRN, Form 499 Filer ID, OCN, STI-PA SPC token, traffic-exchange agreements, and the RMD filing — and observes they “do not currently provide sufficient substantive verification to keep bad actors out of the voice network.” Numeracle asks the Commission to “expressly seek comment on where the verification function should sit, what qualifies an entity to perform it, and how information should be verified, shared, and relied upon.” That’s the KYUP question stated plainly — and Rebekah Johnson sat two seats down from me on the consumer-protection panel two days earlier, a reminder that the trust and transition debates share their cast.

The American Bankers Association wrote directly to Chairman Carr backing the item, framing the RMD as central to “the Commission’s efforts to prevent criminals from placing calls to consumers that impersonate banks,” and pressing the same revolving-door fix Numeracle raised: shut out providers that “appear to exist primarily to facilitate unlawful calling campaigns,” and stop bad actors from reentering “under a new corporate identity.” When the banks and the identity vendors ask for the same thing — real entry-point screening and a durable way to keep expelled actors out — the FNPRM has a constituency well beyond the carriers.

The KYC reply round: ICLE draws the privacy line, Mutare moves the frame upstream

Running alongside the RMD item is the reply round on the KYC proposal proper — the Advanced Methods to Target and Eliminate Unlawful Robocalls FNPRM, FCC 26-32 — where opening comments landed June 25 and replies are arriving now. The International Center for Law & Economics filed the reply worth reading, arguing for “a flexible performance standard rather than a rigid compliance code” because “a rule specific enough to audit would also be static enough to game.” Its privacy argument is the one that matters for identity architecture: prescriptive KYC “would turn providers and their retail channels into high-value targets for data theft,” concentrating “irreversible identity-theft risks on law-abiding consumers, while determined bad actors could still rely on stolen or synthetic identities.” That last clause captures the whole asymmetry of document-based KYC — real collection-and-retention risk on everyone who complies, and the fraudsters route around it with synthetic identities. ICLE also warns that rigid credentialing “would deny or delay essential service” for students, international visitors, domestic-violence survivors, and prepaid users, and notes the intercarrier-comp and IP-transition reforms are already draining the economics out of junk traffic — “under bill-and-keep, the marginal revenue from an illegal call approaches zero.” That’s the same bill-and-keep lever Lumen invoked in the interconnection docket, pointed at fraud rather than at revenue distortion.

The comment from Mutare moves the frame, and it’s the closest thing to an AI-and-voice-fraud thread this week even without reaching for the buzzwords. Its argument: STIR/SHAKEN, the RMD, and KYC/KYUP sit at the wrong layer for the threat that’s actually growing — social engineering. Citing the Google Cloud M-Trends 2026 report that voice phishing is now the second most common initial infection vector observed worldwide, and the leading one for cloud compromises at 23%, Mutare reframes: “Sometimes voice conversation itself is the fraud.” Attackers “exploit human trust before organizations have an opportunity to evaluate communications risk,” increasingly combining voice “with email, SMS, collaboration platforms, remote access software, and artificial intelligence.” Authentication tells you the call was signed; it doesn’t tell you the human on the line is running a help-desk-reset scam. The underlying point — that caller authentication and human-trust exploitation are different problems needing different controls — is a useful corrective to a record that mostly assumes solving origination-identity solves fraud.

Honorable mentions

The onshoring NPRM (CG 26-52, also 17-59, 02-278, 22-2) drew its most organized industry response of the cycle: the Enterprise Communications Advocacy Coalition met with CGB staff July 13 to argue the record “does not support the NPRM’s apparent premise that offshore customer-service operations inherently result in poorer customer experiences, increased fraud, or heightened risks to consumers,” with roughly 42% of surveyed respondents outsourcing customer service and the claim that “geography is a poor proxy for customer-service quality.” The individual-commenter wave in 17-59 and 02-278 kept rolling — a couple dozen filers, mostly without attached comments, plus banking-side voices like the Kansas Bankers Association. And on the national-security side, the Public Safety and Homeland Security Bureau and OET sought comment (DA-26-742, PS Docket 26-184) on prohibiting the import and marketing of previously-authorized foreign-produced drones and UAS components from the entities that drew last week’s enforcement fines — an equipment-supply-chain matter more than a signaling one, but the next procedural step in the same Covered-List push.

Looking ahead

July 22 is the day: watch whether the RMD FNPRM’s “keep expelled bad actors and their principals out” provisions survive into the adopted notice intact, since that revolving-door problem is what both Numeracle and the ABA put at the center of their asks — and whether the adopted item makes good on aligning RMD screening with KYC. On the transition, the interconnection record is converging on the Lumen/USTelecom “light-touch, two-points-of-interconnection” model against Inteliquent’s tariffed NTP framework, with a proposed 2029 forbearance date from legacy TDM rules as the anchor; expect the reply cycle to turn on 911 continuity and the rural cost-recovery concerns WTA raised, with WTA’s own High-Cost comments due in early August. Any recording or transcript the Bureau posts from the workshop’s 911 and consumer-protection panels will be worth reading for how the accessibility and emergency-calling obligations get carried across the cutover. And the KYC reply round in 17-97/17-59 has a little more to run before the record closes.