Contribution · Application — Government

AI for Public Records and FOIA/RTI Triage

RTI (India) and FOIA (US) backlogs run years at some agencies. LLMs can triage requests, search repositories for responsive records, suggest redactions per statute, and draft responses. The tradeoff: faster responses vs. automated over-redaction defeating transparency laws. Design for accountability — every redaction human-confirmed, every exemption cited, every rejection appealable.

Application facts

Domain
Government
Subdomain
Transparency
Example stack
Claude Sonnet 4.7 for triage + redaction suggestion · LlamaIndex over records repositories · pgvector for semantic responsiveness scoring · Redaction UI with officer confirmation · Audit log + appeal workflow

Data & infrastructure needs

  • Records repository (with proper classification)
  • Statutory exemption schedule
  • Historical requests + decisions
  • Requester contact + consent data

Risks & considerations

  • Over-redaction defeating transparency laws
  • Under-redaction leaking classified or protected info
  • Bias — systematically harder responses for marginalized communities
  • Compliance — RTI/FOIA statutory deadlines and appeals
  • Security — classified or sensitive data exposed to cloud LLMs

Frequently asked questions

Is AI for FOIA/RTI triage safe?

With human-confirmed redactions and appealable rationale: yes, and it materially reduces backlog. For classified or highly sensitive agencies, deploy on-premise or sovereign cloud. The LLM assists; the human signs.

What LLM is best?

Claude Sonnet 4.7 with its nuanced handling of exemption reasoning. For high-volume agencies, fine-tune on past decisions to improve consistency. Always deploy where records can legally reside.

Regulatory concerns?

India: RTI Act 2005, DPDPA, Official Secrets Act for exempted info. US: FOIA 5 USC 552, state sunshine laws. EU: PSI Directive / Open Data Directive, GDPR. Statutory deadlines and appeal rights cannot be compromised by automation.

Sources

  1. RTI Act 2005 — accessed 2026-04-20
  2. FOIA — US Archives — accessed 2026-04-20

Related

  • AI for Sanctions and AML Screening — Banks use LLMs plus entity-matching to screen customers, transactions, and counterparties against sanctions lists — reducing false positives and speeding alert triage while staying within FATF/PMLA bounds.
  • AI for Grant Proposal Evaluation — Grant-making agencies use LLMs to summarize proposals, check completeness, detect duplication, and draft reviewer notes — with peer reviewers and program officers making final decisions.
  • AI for Government Permit Processing — Government agencies use document AI and LLMs to triage permit applications, extract fields, check completeness, and draft decisions — speeding service delivery while preserving due process.
  • All Applications — Browse the complete applications library
  • VSET — Programmes offered — B.Tech programs (CSE, AI&ML, AI&DS, Electronics, IIoT)