Methodology
What this site does
Newlaws.in maps every section of the Indian Penal Code, 1860 (IPC), Code of Criminal Procedure, 1973 (CrPC), and Indian Evidence Act, 1872 (IEA) to its corresponding provision in the Bharatiya Nyaya Sanhita, 2023 (BNS), Bharatiya Nagarik Suraksha Sanhita, 2023 (BNSS), and Bharatiya Sakshya Adhiniyam, 2023 (BSA). For each pair, the page surfaces:
- The verbatim old-section and new-section text (where text-verified)
- A relationship classification: substantively same, modified, partial, or no correspondence
- Structured uncertainty flags (scope drift, definition drift, consolidation context, etc.)
- A change-note describing the textual delta
- Field-level confidence labels for every claim
- Source links and discussion-log references
What this site does not do
- It does not interpret the law.
- It does not give legal advice.
- It does not predict how courts will rule.
- It is not a substitute for a qualified advocate.
If a claim cannot be pointed to a specific line in the statute or a linked source, it does not get published.
Sources used
- India Code (https://www.indiacode.nic.in/) for the IPC, CrPC, IEA bare acts
- Gazette of India PDFs for the BNS (Act 45 of 2023), BNSS (Act 46 of 2023), and BSA (Act 47 of 2023)
- PRS Legislative Research correspondence charts for old-to-new mapping baseline
- Supreme Court of India judgments cited by name where directly relevant
Mapping process
- AI extraction. The agent reads the bare act PDF for the new code and extracts each section's verbatim text into a structured ActSection record.
- Cross-check protocol. Each draft mapping is run through two independent LLM model runs on the same prompt and inputs. The harness diffs the canonical fields (relationship, coverage, from/to id, input coverage). Agreement promotes the edge to
cross_checked; disagreement keeps it atai_draftwith a structured disagreement note. - Editor verification. A human verifier text-checks the section text on both sides against the bare act and promotes the edge to
editor_verified. - Publish gate. Once schema validation, cross-reference integrity, prose constraints, and HTML allow-list checks pass, the page becomes eligible for
published.
What the labels mean
- Text-verified — the text is taken directly from the statutory source (bare act PDF or India Code).
- AI-indicated (source-linked) — the mapping has been extracted from the schedule, source, or correspondence chart. Source link is provided.
- AI-inferred — derived from textual structure or context (used for FAQs, examples, common-misconception sections).
- Uncertain — verify — there are conflicting signals or insufficient evidence. Verify against the official source before relying.
- Prior ruling noted; application under new law may vary — applies to case-law continuity claims. Pre-2024 judgments retain interpretive value but their application to BNS / BNSS / BSA depends on judicial development.
Schedule handling
The First Schedule (bail / cognizable / compoundability classifications) is parsed on-touch as pages go live, not in bulk. Any high-risk field carries a visible field-level label and a “verify against the latest schedule” cue.
Case-law handling
Where a Supreme Court ruling directly informs a section's interpretation (e.g., Mithu v. State of Punjab for IPC 303 → BNS 104, or Joseph Shine v. Union of India for IPC 497), the case is named and the relevant 25-word holding is cited verbatim. The site does not extrapolate beyond what the holding says.
Corrections policy
See the corrections page. User-submitted corrections are reviewed and credited.
Update frequency
The corpus is updated as the agent finds new authoritative sources, as users submit corrections, and as the editor processes the verification queue. Each page records its last_updated date.