On October 10, 2026, at 09:00:39 UTC, the federal docket TREAS-DO-2026-0496 registered its final modification. Four documents. One frozen timestamp. No amendment since. And yet, on compliance desks from New York to Singapore, a different date keeps circulating: November 4. That date is a ghost. It belongs to a different year, a different rulemaking, and a comment period that closed eleven months ago. File against it, and the Treasury's electronic docket will reject your submission automatically. No human reviewer. No courtesy email. No appeal.
I spent the last three days pulling the primary source. Here is what the record actually says, and why an entire industry nearly missed its only structured shot at shaping the rule.
The GENIUS Act โ Public Law 119-27 โ is the first federal statute to carve out "payment stablecoin" as its own regulatory category, separate from securities and separate from bank deposits. That classification matters more than any token launch this cycle. It decides who can issue a dollar-pegged instrument, who can distribute it, what reserves must back it, and when the whole machine has to be compliant.
The statute delegates the detail work to three federal bodies. Treasury owns the issuance and reserve framework. The Office of the Comptroller of the Currency handles bank-grade infrastructure and custody. The Federal Reserve runs a parallel rulemaking for the clearing and settlement layer. Three agencies, three rulebooks, one shared procedural clock. A prior piece in this series framed it cleanly: "Two NPRMs, Two Paths, One Deadline."
Treasury's Notice of Proposed Rulemaking dropped on August 18, 2026. Standard Administrative Procedure Act mechanics: a 60-day public comment window. Sixty days from August 18 lands on October 17 โ a Saturday. Under federal practice, a deadline falling on a weekend rolls to the next business day. That gives you Monday, October 19, 2026, 23:59:59 Eastern Time.
That is the real deadline. It is not ambiguous, not under negotiation, and locked by the docket's own configuration, which is set to reject late submissions outright.
So where does November 4 come from? It comes from 2025. Last year's Advance Notice of Proposed Rulemaking โ a separate, earlier procedural step โ carried its own comment deadline, and that historical notice is cited inside the current record at footnote 16. Footnote 16 is a reference, not an instruction. But automated retrieval systems do not read footnotes the way lawyers do. They read them as content. And here is the trap: the 2025 deadline was October 20. The current deadline is October 19. One year apart. Two days apart on the calendar. Same month, near-identical day, identical subject matter.
I call this a calendar collision, and it is the single most dangerous failure mode in automated regulatory compliance. A retrieval tool keyed on "GENIUS Act comment deadline" will surface the 2025 footnote happily, because the text similarity is nearly perfect. The only distinguishing feature โ the year โ is exactly the token an embedding-based search blurs hardest.
I verified the current docket directly. TREAS-DO-2026-0496 holds exactly four documents. The final modification timestamp is October 10, 2026, 09:00:39 UTC. The comment-period field has not moved. The system refuses submissions after the cutoff. That is the entire evidentiary chain, and anyone with docket access can reproduce it.
Here is what should worry you more than the date itself. Sixty-five comments have been filed. That is the total public participation for a rule that will restructure a market measured in the hundreds of billions. The distribution is worse than the number. Thirty-six of those sixty-five โ roughly 55% โ arrived in the final fifteen days. Nine of them โ about 14% โ landed in the last week. A textbook deadline-sprint curve. The industry, collectively, did not treat this as a priority until the clock was nearly out.
Now overlay the ghost footnote. A compliance team that discovered the rule late, rushed a submission, and leaned on a search engine to confirm the deadline would have targeted November 4. They would have filed nothing on October 19. They would have learned about the error only when the docket refused them โ after the window had sealed.
That is not a hypothetical inefficiency. It is a structural advantage for whoever filed early and correctly. If part of the market misses the window because of a misread date, their demands never enter the formal record. The rule then tilts toward the positions that were actually submitted. First-mover advantage, inside a rulemaking.
Let me be precise about what November 4 is not. It is not an extension. Not a grace period. Not a Treasury accommodation. It is a recycled citation from a document that concluded its own comment process a year ago. The current record contains no extension language of any kind.
I have to flag one weakness in my own chain, because forensic work demands it. The broader effective-date timeline rests on Section 20(a) of Public Law 119-27, which fixes the statute's activation as the earlier of two milestones: eighteen months after enactment, or 120 days after final rules are published. My reading comes from two consistent retrieval passes and Treasury's own NPRM statements โ not from the statute text itself. Congress.gov returned a 403 during my checks, and the govinfo page was unavailable. If Treasury's interpretation of Section 20(a) is off, the downstream calendar shifts. Mark that as an open thread, not a settled fact.
With that caveat, the projected effective date is January 18, 2027. And the headline prohibition โ proposed rule 1523.3(a), barring digital asset service providers from offering and selling payment stablecoins โ is set for July 18, 2028. Distributors get a three-year runway, but the runway starts when the rule lands, not when they start preparing.
The economics underneath are worth stating plainly, because the date debate hides them. A payment stablecoin under this framework is a reserve-backed instrument, and the reserve rules determine the issuer's entire margin. Tighter reserve definitions raise the cost of holding the backing. Broader distribution bans reshape who can even touch the product. The comment window is the last point at which an issuer can argue those definitions before they harden into a final rule โ and a final rule is far harder to litigate than to comment on.
Here is the contrarian read the panic posts are missing. Everyone is treating this as a story about a deadline. It is not. It is a story about the reliability of automated legal research, and the stablecoin window is just the first place it drew blood in public.
The footnote format โ "see 90 FR 47251" โ is engineered for human citation practice. To a keyword index or an embedding retriever, that string is not a pointer. It is content. The system cannot distinguish "here is the current rule" from "here is a historical rule we are referencing." So it surfaces both, weighted by textual similarity, and the year โ the one token that disambiguates โ is the token it handles worst.
This is the same failure class I watched in on-chain forensics during the FTX unwind, when wallet-clustering tools kept merging distinct entities because their heuristics over-weighted transfer proximity. The tool was never lying. It was answering a slightly different question than the one being asked. Regulatory retrieval tools do the same. They answer "what text mentions this deadline" when you need "what is the operative deadline." Those are not the same query, and the gap between them is where filings die.
There is a second blind spot nobody is discussing. Sixty-five comments is a suspiciously thin record for a rule of this magnitude. Three explanations fit. One: the industry is genuinely watching and waiting. Two: participants route their positions through trade associations, betting a single unified voice beats sixty-five scattered ones. Three: most firms simply have not grasped how much this rule will cost them. Any is defensible. But combined with the sprint curve, the likeliest reading is that the industry under-weighted this proceeding until very late โ and a late start is exactly when a ghost footnote does its damage. The participation data and the misread risk are not two problems. They are one problem wearing two masks.
One more structural note that the coverage has skipped entirely. Payment stablecoin regulation has always collided with state money-transmitter licensing, and that federal-versus-state tension is absent from the current debate. The omission matters, because a rule that settles the federal layer without addressing the state layer just relocates the compliance burden rather than removing it.
Watch three things from here. First, whether Treasury's final rule text preserves the October 19 record or reopens any window โ a reopening would signal the misread was widespread enough to force a correction. Second, the OCC's November target, which the agency has described as a "labeled claim" rather than a fixed regulatory fact. If OCC moves before Treasury finalizes, the bank-side plumbing lands ahead of the issuance rules, and that sequencing gap becomes its own puzzle. Third, whether anyone who missed the window attempts a "reasonable reliance" defense โ arguing they were misled by an official source. Proving that is brutal, and it requires showing the misread originated from a government docket rather than a private search tool.
The docket does not care that the search engine was confident. October 19 is the date. November 4 is a ghost. The only question left is how many filing calendars still have the wrong one printed on them.