Six selected excerpts come from Apple Inc.'s original FY2024 and FY2025 Form 10-K HTML files. This is a convenience example for narrative comparison, not a representative or held-out dataset. Prepared October 13, 2026.
Original filings
| Fiscal year | Period end | Accession | Primary filing |
|---|---|---|---|
| 2024 | 2024-09-28 | 0000320193-24-000123 | Original SEC HTML |
| 2025 | 2025-09-27 | 0000320193-25-000079 | Original SEC HTML |
The source text belongs to the issuer's disclosures. This project does not claim authorship of that text or grant rights to third-party material.
The supplied HTML copies were read locally and checked against these SHA-256 values before selection:
- FY2024:
a051b50fcba8c752df3dd55a87f42e0fb1c8f31b7b8ed287f55c8c385849e159 - FY2025:
274361a794f84b24d75b4ee98bc8eff5a031b4fc1be1a1210dbc2b10d92c5179
These hashes identify the inspected local copies. They do not establish byte-for-byte identity with files fetched independently from SEC. The local FY2024 copy contains 1,503,779 bytes; the SEC index lists 1,503,780 bytes. The local FY2025 copy contains 1,520,207 bytes; the SEC index lists 1,520,208 bytes. These one-byte discrepancies are unresolved, and the source hashes must not be presented as authenticated SEC-hosted hashes. The public program reads only the bundled excerpts and does not fetch or rehash complete source files.
Selection and transformations
Both filings use Item 1A, Risk Factors. Within the 2024 source, the relevant anchor prefix is i7bfbfbe54b9647b1b4ba4ff4e0aba09d; within 2025 it is i719388195b384d85a4e238ad88eba90a.
| Pair | Locator and selection |
|---|---|
ip |
Anchor suffix _58, heading “The Company relies on access to third-party intellectual property…”; complete following paragraph. |
ai_warning |
Anchor suffix _58, product-design/manufacturing-defects risk heading; exactly one sentence beginning “The introduction of new and complex technologies…” from the paragraph beginning “The Company offers complex hardware and software products and services…” |
privacy |
Anchor suffix _61, privacy/personal-information risk discussion; complete paragraph beginning “The Company makes statements about its use and disclosure…” |
Full anchors and paragraph prefixes are recorded per excerpt. These anchors locate a broader block; use the paragraph prefix as well as the anchor to find the selected passage. Corresponding headings and paragraph content were reviewed across years, rather than pairing by position alone.
Preparation decoded HTML entities, joined inline text in document order, and normalized whitespace. The packet preserves the resulting selected language, including the 2024 licensing qualification. No model generated or rewrote the excerpts. The preparer selects known unique paragraph prefixes and checks source hashes; it is not a general filing parser. The AI-warning control deliberately omits the rest of its parent paragraph, which contains other changes.
Data dictionary
| Field | Meaning |
|---|---|
schema_version |
Packet version, currently 1. |
records |
Six excerpts: earlier/later crossed with IP, AI warning, and privacy. |
id, kind |
Stable excerpt identifier and paired passage category. |
company, cik, form, accession |
Recorded issuer and original filing identifiers. |
fiscal_year, period_end |
Fiscal year and reporting period end, YYYYMMDD; these are not release or trading dates. |
source_file, source_url |
Original filename and public SEC source link. |
source_sha256, source_size_bytes |
Hash and size of the locally inspected complete HTML copy. |
section, anchor, paragraph_prefix, selection |
Location and scope of the selected text. |
text, excerpt_sha256 |
Selected normalized text and SHA-256 of its UTF-8 encoding. |
Hash checks detect disagreement with recorded excerpt text. Metadata checks detect internal inconsistencies. Neither proves source authenticity, completeness, business meaning, or the truth of a company statement. A changed sentence is evidence of changed disclosure; it is not automatically evidence of a newly emerging risk or investment value.