In May 2026 the SEC proposed to simplify public-company filer status and broaden emerging-growth-company accommodations (Release Nos. 33-11419; 34-105515, File No. S7-2026-18). The proposal would raise the large-accelerated-filer public-float threshold from $700 million to $2 billion, collapse today’s five overlapping filer categories into two (LAF and NAF), and extend scaled disclosure — including exemption from the §404(b) auditor attestation on internal controls — to a much larger population of companies. From the date of publication, the public has 60 days to comment; the deadline is July 20, 2026. Read the proposing release (PDF), or browse the SEC’s public comment file.
This tracker was prepared by Tzachi Zach as a public service, using independent ChatGPT and Claude classification ensembles — to log the comment letters as they arrive, classify each letter’s stance on every element of the proposal it engages, and surface the patterns in the docket. The classification choices and comparison rules are explained in the Methodology accordion below.
This project is part of a series. It applies the same classification approach as my tracker for the SEC’s related semiannual-reporting proposal (File No. S7-2026-15) — a much larger docket that drew tens of thousands of comment letters — which is available here and is worth a look for context on how these dockets unfold. New to filer status? Open “What each element means” for a plain-language guide to A1–C1 and GEN. Comments, suggestions, or corrections welcome.
Classification providers. Every letter is classified independently by ChatGPT and Claude. Each provider uses its own three-rater ensemble: a balanced Primary, a strict Literalist, and a cautious Skeptic for element stance; commenter type uses parallel Primary, Self-described, and Letterhead/functional perspectives. An element enters a provider’s assessment when at least two raters identify it as engaged, and its stance is the resulting two-rater or unanimous majority. ChatGPT and Claude results are stored separately and are never silently averaged or substituted for one another.
Provider views. ChatGPT and Claude views derive every chart and table solely from the selected provider. Compare view places both providers side by side in the overall, proposal-element, and theme charts, using identical scales and denominators. Each element tooltip shows the three underlying votes, evidence, and whether agreement was unanimous or two of three.
How an overall label is computed. Every letter is first scored on each proposal element it engages as Support, Oppose, or Conditional. A provider’s single overall label is then computed from that provider’s element-level positions:
GEN and six-vote comparison. GEN is an assessed reading of a letter’s stated whole-proposal direction, drawn from general-direction passages; it is one input to the computed overall label, not the overall label itself. Compare tooltips also report an experimental six-vote reading across the three ChatGPT and three Claude stance votes. It is descriptive only and never replaces either provider or drives the principal charts.
Historical completeness. Some early classifications preserved the provider-level entity majority and agreement but not all three individual entity votes. Missing historical entity votes remain explicitly unavailable and are never reconstructed by inference.
Scope and status. Letters whose real subject is another docket are relocated, not scored — which is why #16 and #23 are intentionally vacant. This is a living tracker, current as of the date shown at the top; the comment period is open through July 20, 2026.