Why this translation exists

Every major modern English Bible is the product of a closed process. A committee of scholars makes thousands of translation decisions behind closed doors; readers receive the finished text; the reasoning behind any given word choice — why doulos is rendered "servant" while its bonded-status nuance remains explicit in notes and the public audit trail, why the footnote elevates one reading over another — is almost never surfaced to the people actually reading scripture.

This worked for centuries because it had to. Assembling the scholarship for a Bible translation required publishers, institutional funding, and closed-room deliberation. The opacity wasn't a choice — it was the cost of getting the work done at all.

It isn't any longer.

The moment this becomes possible

Three things are true now that were not true a decade ago:

Open source infrastructure — public repositories, issue trackers, collaborative editing — is free, instant, and globally accessible. A reader anywhere in the world can audit a translation decision the moment it's committed.
Frontier AI models — Claude, GPT, Gemini — can produce competent drafts of biblical Greek and Hebrew translation with full lexical reasoning, exposing every decision they make.
Modern provenance standards — versioned source text, prompts where preserved, model metadata, decisions, and published editions let third parties audit what was done. Hosted AI models are not guaranteed to return identical text when rerun, so provenance is not a claim of byte-for-byte AI reproducibility.

Together, these make something new possible: a translation where every decision is documented, every disagreement is public, every verse is inspectable, and every word can be traced back to the Greek or Hebrew it came from — in under 60 seconds, from a phone, anywhere in the world.

That is the People's Open Bible.

What we are translating

We translate directly from public-domain and openly-licensed scholarly editions of the original-language texts, across all three canonical sections of the Christian Bible.

New Testament (27 books):

SBLGNT (Greek New Testament, ed. Michael W. Holmes) — the closest legally-available approximation to the autographs the NT authors wrote.

Old Testament — Protestant canon (39 books):

Westminster Leningrad Codex and unfoldingWord Hebrew Bible — based on the Leningrad Codex (1008 AD), the oldest complete Hebrew Bible manuscript in existence.

Deuterocanonical / Apocrypha (14 books — Tobit, Judith, Wisdom of Solomon, Sirach, Baruch, Letter of Jeremiah, Additions to Esther, Additions to Daniel, 1 Maccabees, 2 Maccabees, 3 Maccabees, 4 Maccabees, 1 Esdras, Prayer of Manasseh, Psalm 151. See DEUTEROCANONICAL.md for why we include these and which Christian traditions receive each as canonical):

Swete LXX (Henry Barclay Swete, The Old Testament in Greek According to the Septuagint, Cambridge, 1909–1930) — a public-domain diplomatic edition of Codex Vaticanus. This is our Greek source for every deuterocanonical book. We transcribe it ourselves from the archival page scans using AI vision and release the transcription under CC-BY 4.0.
Sefaria Ben Sira (Kahana edition, CC0) and Schechter 1899 — the public-domain Cairo Geniza Hebrew witness to Sirach. Roughly two-thirds of Sirach survives in Hebrew; we translate from the Hebrew where it survives and from the Greek where it does not.
Neubauer 1878 Hebrew Tobit (via Sefaria, Public Domain) — a Hebrew back-translation of Tobit, used as a Semitic-phrasing reference alongside the Greek Long Recension (Codex Sinaiticus).
WLC MT parallels for 1 Esdras — 1 Esdras is a Greek recomposition of 2 Chronicles 35–36, Ezra, and Nehemiah 7–8. We cross-reference each 1 Esdras verse to its Hebrew parallel in our WLC corpus, with the "Story of the Three Youths" (3:1–5:6) treated as Greek-only material with no Hebrew parallel.

We are not paraphrasing or smoothing an existing English translation. We are translating from the same sources the scholarly community works from — the Greek and Hebrew — applying documented translation philosophy to each verse, with every decision auditable.

What "transparent" actually means here

Every verse in this translation is a file in this repository. Each file contains:

The source text (Greek or Hebrew) being translated
The English rendering that ships to readers
Every lexical decision made — the source word, the chosen English gloss, the alternatives considered, the lexicon consulted, and why the chosen gloss was preferred
Every theologically-contested reading, with the alternative preserved in footnotes rather than buried
The AI model, prompt hash, and timestamp of the draft
A git commit history documenting every revision

If you disagree with a rendering, you don't have to send a letter to a publisher and hope for a reply. You open an issue on GitHub, cite the specific verse, and engage publicly. We commit to responding substantively to every serious concern. When we revise, the revision is itself a commit with documented rationale. Nothing happens in private.

This is transparency in the same sense scientific papers are transparent: we show the data, we show the methods, we show the reasoning, and we invite the rest of the world to check our work.

Current status

The project has reached near-complete first-draft coverage of the Christian biblical corpus, with a multi-model revision pipeline running on top of the original drafting work. Live numbers are at the translation progress page; as of the most recent snapshot:

Protestant New Testament: all 27 books, 260 chapters, ~7,900 verses.
Protestant Old Testament: 38 of 39 books drafted (every book except Song of Songs), ~23,000 verses across both Torah/Histories/Wisdom and Prophets.
Deuterocanonical / Apocrypha: all 18 books drafted (Tobit, Judith, Wisdom of Solomon, Sirach, Baruch, Letter of Jeremiah, Greek additions to Esther and Daniel, 1–4 Maccabees, 1 Esdras, Prayer of Manasseh, Psalm 151) — over 6,000 verses, translated from Swete LXX with the Hebrew witnesses for Sirach and Tobit consulted in parallel.
Extra-canonical witnesses translated for transparency about the Jewish/early-Christian textual world (Didache, 1 Clement, 1 Enoch, Jubilees, Psalms of Solomon, 2 Esdras) — not claimed as scripture, but surfaced so readers can see what existed alongside the canonical books.

Every verse remains a draft. What has changed since the project began is the depth of the review pipeline, not the claim about finality:

Drafting: a frontier drafter model produces the first English rendering for each verse, anchored in the doctrine prompt and the verse's source-language text (Greek, Hebrew, or Aramaic depending on the book).
Revision pass: Gemini 3.1 Pro reads each draft against the source text, identifies lexical disagreements, awkward English, and category-1 grammar issues, and adjudicates whether to apply a change or mark the verse unchanged. Every revision is recorded in the verse YAML's revisions array with from, to, rationale, and a model attribution — so anyone reading the file can see the full negotiation history.
Regression checks: a regression-fix layer catches cases where an automated revision silently violates a documented project policy (e.g., Χριστός being changed back to "Christ" against DOCTRINE.md). Those reverts are also visible in the same revisions array.
Direct human review (still informal): commits land in main through standard git workflow with public diffs. Issue tracker is open; every revision can be challenged.

This is no longer "what one AI produced." It is a drafter, an independent revision pass on a different model family, and a policy audit — with the entire negotiation persisted in version control.

What it is not yet: signed off by a credentialed scholar with named authority over the rendering. We do not pretend to that, and the commit history makes the absence visible. When credentialed-scholar review becomes part of the process, it will be announced publicly and documented in the repository in the same auditable way.

The text is also shipping live to readers through the Cartha mobile app and the web reader, through a CDN publisher pipeline that pushes whatever is on main. The app, the website, and this repository show the same draft, with the same provenance, at all times.

Open by principle, not by default

We release this translation under Creative Commons Attribution 4.0 International (CC-BY 4.0) — the canonical open-content license. This means:

Anyone may use it — in apps, websites, books, sermons, study curriculum, audio Bibles, video content, academic papers.
Anyone may use it commercially — publish a paid study Bible, sell an app, build a paid tool. No royalties flow to us. No permission required.
Anyone may create derivative works — paraphrases, adaptations, commentaries, translations into other languages, revisions.
Anyone may redistribute it, including modified versions.
Attribution is the only requirement: credit the People's Open Bible, link to the license, and indicate if you made changes.

This is a deliberate theological choice, not a pragmatic one. Every major Bible before 1900 was effectively public domain — the KJV, Geneva, Tyndale, Luther's German, Wycliffe. Paid licensing of scripture is a modern innovation, not a historical norm. Our commitment is to recover the older pattern: freely you have received, freely give (Matthew 10:8).

We do not want to own a Bible. We want to help produce one, hand it to the church, and have people read it.

What this is not

Several things it helps to say explicitly:

Not a finished translation. What's in the repository is a draft. It will have errors, awkward renderings, and passages that will need revision. Every release tag makes this explicit.
Not a replacement for scholarship. We stand on the shoulders of centuries of Greek and Hebrew lexicography, textual criticism, and theological scholarship. Every lexical decision cites the lexicons (BDAG, HALOT, LSJ, Louw-Nida) the academy already uses.
Not a denominational product. POB does not commit to a specific denominational or creedal stance. Where the source texts presuppose theological claims, the translation reflects those claims as the texts present them, not as imposed commitments. Contested readings are preserved in footnotes, not smoothed away. See DOCTRINE.md for the full translation-stance document.
Not a study Bible. Footnotes document translation decisions, not devotional application.
Not finished. This is an ongoing project. We ship phase-by-phase, book-by-book, with every stage transparently marked as draft.

How to engage

Read it. When a book is drafted, read it. Tell us what lands and what doesn't.
Critique it. File an issue on GitHub for any verse you'd translate differently. Lexical disagreements, theological disagreements, and general concerns all have templates.
Cite it. In academic work, sermons, or published material, cite the People's Open Bible with a link back. Attribution keeps the paper trail visible.
Fork it. If you have a specific scholarly disagreement you want to pursue, fork the repository and publish your alternative. Our provenance records travel with the fork.
Translate it. CC-BY 4.0 permits derivative translations into other languages. A translator working from the People's Open Bible into any other language inherits the full provenance chain back to the Greek and Hebrew.

Our commitment

We commit to:

Never paywall scripture. The text will always be free to read.
Never hide decisions. Translation choices remain documented and publicly inspectable; stable edition artifacts can be verified by checksum.
Never silence disagreement. Public issues remain open; responses are public; revisions are public.
Never claim authority the text doesn't have. This translation is an AI-produced rendering of ancient documents into modern English, released openly. It is not itself inspired. Scripture is.

The Word belongs to the church. We are stewards of a process, not owners of a text.