NARIT-AI: When Algorithms Become Investigators — Gujarat's AI Gambit for NDPS Justice

Introduction & Context: AI at the Intersection of Justice & Security

📖 Introduction — NARIT-AI

What Is NARIT-AI?

NARIT-AI — Narcotics Analysis & RAG-based Investigation Tool — is a specialized, closed-environment artificial intelligence system developed by the Gujarat Police in collaboration with Gradiante Creative Services (a Mumbai-based AI startup) and the Western Railway Police, Vadodara Division. Launched on 10 April 2026 in Gandhinagar, it is officially the first AI-based investigation tool deployed by any state police force in India, marking a paradigm shift in how law enforcement interacts with legal knowledge and evidence analysis.

The tool is purpose-built to address a precise, measurable failure in the criminal justice system: the recurring acquittal of accused persons in NDPS (narcotics) cases not because of insufficient evidence, but due to procedural lapses by investigating officers during search, seizure, and documentation. These lapses — non-compliance with Sections 41, 42, 50, and 52A of the NDPS Act — have consistently allowed guilty parties to walk free, eroding public trust in the law enforcement apparatus and creating a perverse incentive for narcotics networks to exploit legal technicalities.

Why Does It Matter — Right Now?

NARIT-AI emerges at a critical juncture in India's internal security narrative. The drug trafficking threat has intensified — India sits geographically between the Golden Crescent (Afghanistan, Iran, Pakistan) and the Golden Triangle (Myanmar, Thailand, Laos), making it both a transit corridor and an increasingly significant consumer market. The total value of narcotics seized across India in 2024 surged roughly 55% year-on-year to approximately ₹25,330 crore. Simultaneously, India's criminal justice infrastructure is grappling with the transition to three new criminal laws — Bharatiya Nyaya Sanhita (BNS), Bharatiya Nagarik Suraksha Sanhita (BNSS), and Bharatiya Sakshya Adhiniyam (BSA) — that have elevated procedural compliance requirements. In this context, NARIT-AI is not merely a policing tool; it is a test case for whether AI can be responsibly deployed in high-stakes governance functions.

📌 Origin of the Name

NARIT-AI = Narcotics Analysis & RAG-based Investigation Tool — Artificial Intelligence. The acronym deliberately centres RAG (Retrieval Augmented Generation) as its defining technical identity.

✍ Mains Tip

In any Mains answer, open with the problem NARIT-AI solves, not the tool itself. The hook is: "Despite strong evidence, a conviction rate of 25% in NDPS cases (2022) reveals a systemic procedural deficit — one that NARIT-AI attempts to address through algorithmic compliance engineering." This framing immediately signals analytical depth.

⚠ The Problem Before NARIT-AI

Only specially trained officers could handle NDPS cases
Procedural lapses leading to acquittals despite evidence
Gujarat NDPS conviction rate: ~25% (2022)
Workload concentration creating investigative bottlenecks
Uneven legal knowledge across police ranks and districts
No real-time guidance during field investigations

✅ What NARIT-AI Offers

Any-rank officer can conduct quality NDPS investigations
Procedural checklist generated from FIR content automatically
Real-time legal guidance trained on SC/HC judgments
Evidence of equal quality across rural and metro officers
Anticipates defence arguments and prepares rebuttals
Force multiplier: expands specialist capacity without specialist staff

NARIT-AI is not just a technology intervention — it is an institutional response to a crisis of procedural compliance that had systematically undermined India's war on narcotics. Its significance extends far beyond Gujarat.

India's Drug Menace & Legal Framework: The Context That Created NARIT-AI

📖 Introduction — Contextual Foundation

India's Geographic Vulnerability: Sandwiched Between Death Zones

India's internal security is structurally shaped by its geographic location between Asia's two largest illicit drug-producing zones. The Golden Crescent — comprising Afghanistan, Iran, and Pakistan — historically accounts for the bulk of opium and heroin flowing into India's western and northern states (J&K, Punjab, Himachal Pradesh, Rajasthan, Gujarat). The Golden Triangle — Myanmar, Thailand, Laos — feeds methamphetamine and heroin into India's northeastern states, with Moreh (Manipur) identified as a key drug conduit. This dual vulnerability makes India simultaneously a transit route and a growing consumer market. The drug-terror nexus amplifies the security dimension: narcotics revenues finance insurgencies in the northeast and cross-border terrorism from Pakistan.

₹25,330 cr

Value of drugs seized, India 2024 (↑55% YoY)

80 quintals

Meth-type stimulants seized 2024 (vs 34 in 2023)

5,500 kg

Meth haul from Myanmarese vessel, Nov 2024 — largest in Coast Guard history

25%

NDPS conviction rate in Gujarat, 2022 — the trigger for NARIT-AI

292 mn

Global drug users (UNODC 2022) — 20% rise in a decade

Constitutional Mandate: Article 47 — The Root of India's Anti-Narcotics Policy

India's entire narcotics policy traces its constitutional anchor to Article 47 of the Constitution — a Directive Principle of State Policy — which directs the State to "endeavour to bring about the prohibition of the consumption of intoxicating drinks and of drugs which are injurious to health, except for medicinal purposes." This is the philosophical bedrock upon which the NDPS Act is constructed. While Article 47 is non-justiciable, it has been consistently cited by courts to justify the Act's stringent penal framework and to resist arguments that it disproportionately curtails individual liberty.

NDPS Act 1985 — Legislative Evolution at a Glance
Year / Event	What Changed	Significance
1961	UN Single Convention on Narcotic Drugs — USA campaigns globally for cannabis ban	International pressure triggers India's eventual legislative action
1985	NDPS Act passed (Act 61 of 1985); assented 16 Sep, in force 14 Nov 1985	Comprehensive prohibition: cannabis, opium, psychotropics; non-bailable offences
1986	Narcotics Control Bureau (NCB) constituted under the Act	Central coordination body for drug law enforcement
1988	First amendment — Section 27A added (financing illicit traffic)	Extended reach to narco-terrorism financing
2001	Second amendment — sentencing rationalized, bail liberalized for addicts	Shifted from punitive-only to rehabilitative approach for small offences
2014	Third amendment — Essential Narcotic Drugs defined; mandatory death penalty for repeat offences removed (30-yr sentence as alternative)	Balanced medical access (morphine, fentanyl) with enforcement; addressed drafting error in S.27A
2021	Fourth amendment	Further procedural updates
2024–26	BNS, BNSS, BSA replace IPC, CrPC, Evidence Act	NARIT-AI trained on all three new codes; investigators must now comply with dual legal frameworks

Key NDPS Sections That Drive Investigation Quality (and Acquittals)

The NDPS Act's procedural stringency is both its strength and its Achilles heel. Several sections impose mandatory compliance requirements that, if violated even technically, can vitiate an entire prosecution. This is precisely the gap NARIT-AI targets.

Section 41/42 — Conditions for search and seizure with/without warrant; must record grounds in writing before action
Section 50 — Right of the searched person to demand search before a Gazetted Officer or Magistrate; non-compliance = fatal to conviction
Section 52A — Mandatory procedure for disposal of seized drugs; failure vitiates conviction regardless of drug quantity
Section 67 — Power to call for information; confessions recorded here are inadmissible (per Tofan Singh 2020)
Section 37 — All offences cognizable and non-bailable; bail conditions are strict for commercial quantities

★ Critical Insight for Mains

The NDPS Act is unusual in Indian criminal law: it is a strict-liability procedural statute where the accused need not prove innocence — the prosecution must prove both guilt and procedural compliance. This "double burden" on the State makes NARIT-AI's compliance-engineering approach directly responsive to India's conviction rate crisis.

NARIT-AI is not merely a technological response to a policing problem — it is a constitutional imperative (Art. 47) meeting a criminal justice failure (25% conviction rate), mediated through artificial intelligence. This framing is essential for Mains answer writing.

NARIT-AI: Technical Architecture, Key Features & What Makes It Unique

What Is RAG Technology? — The Engine Beneath NARIT-AI

Retrieval Augmented Generation (RAG) is an AI architecture that combines two processes: first, retrieval — querying a curated, structured knowledge database to find relevant documents, precedents, or provisions; and second, generation — producing a coherent, contextually accurate output by synthesizing the retrieved material. Unlike standard Large Language Models (LLMs) that rely on probabilistic pattern matching from internet-scale training data, a RAG system grounds every output in specific, verified documents from its closed database. This is why NARIT-AI can claim low "hallucination" risk — its knowledge is bounded, auditable, and domain-specific. When an investigating officer uploads an FIR, NARIT-AI retrieves the most relevant NDPS provisions, SC/HC precedents, and procedural guidelines, then generates a structured analysis report tailored to the specific facts of that case.

📌 The Closed Database Advantage

NARIT-AI does not access the open internet. Its knowledge base consists exclusively of: NDPS Act 1985 bare text · BNS, BNSS, BSA bare acts · Landmark SC and HC judgments on NDPS cases · Government circulars and guidelines. This eliminates the hallucination risk that plagues general-purpose AI and ensures legal precision.

NARIT-AI — Step-by-Step Workflow for an Investigating Officer
Step	Action	Output
Step 1	Officer uploads FIR content into NARIT-AI application	System ingests case facts, drug type, quantity, location, circumstances of seizure
Step 2	RAG engine retrieves relevant NDPS sections, case law, procedural guidelines	Legal knowledge base matched to specific case profile
Step 3	AI generates comprehensive investigation analysis report	Case strengths, case weaknesses, evidentiary gaps identified
Step 4	Evidence checklist produced	Mandatory evidence list based on offence type (possession/trafficking/cultivation)
Step 5	Investigation plan and timeline generated	Step-by-step roadmap from seizure to chargesheet
Step 6	Predictive courtroom strategy module	Anticipated defence arguments + suggested prosecution rebuttals based on past judgments
Step 7	Drafting support (chargesheet / court summary)	Draft documents aligned with BNS/BNSS format requirements

What Makes NARIT-AI Structurally Different From General AI Tools

NARIT-AI is classified as a private AI system under the recent guidelines of the Gujarat High Court on AI use — meaning it is restricted exclusively to verified Gujarat Police personnel and carries enforcement-grade double-layer encryption. It is inaccessible to the public, defence counsel, or any external party. This design choice reflects a deliberate trade-off: maximum operational security in exchange for reduced transparency to external scrutiny. The tool is also designed for dynamic updates — its database is meant to auto-update when laws are amended or new judicial circulars are issued, ensuring currency of legal advice.

One of its most analytically significant features is the predictive court strategy module: by learning from patterns in how defence counsel have successfully challenged prosecution evidence in past NDPS trials, NARIT-AI can proactively flag which aspects of a current investigation are most vulnerable to challenge, allowing the investigating officer to plug those gaps before the chargesheet is filed. This represents a qualitative shift from reactive investigation (fixing errors after they are challenged in court) to proactive legal engineering.

RAG Technology

Closed Database

FIR Analysis

Evidence Checklist

Predictive Court Strategy

Draft Chargesheet

Double Encryption

Dynamic Legal Updates

Force Multiplier

NDPS + BNS + BNSS + BSA

✅ Institutional Significance

NARIT-AI is the first application of RAG-based AI to criminal investigation in India. It ensures that a police officer in a remote rural district has access to the same quality of legal expertise as a specialist in a metropolitan crime unit — democratizing investigative capacity across the police hierarchy.

NARIT-AI's RAG architecture makes it legally grounded, hallucination-resistant, and jurisdiction-specific — qualities that distinguish it from generic ChatGPT-style tools and make it an appropriate deployment in high-stakes criminal justice contexts.

Issues & Challenges: The Dark Side of Algorithmic Justice

⚡ Issues — NARIT-AI & AI in Policing

🔍 Critical Analysis — The Accountability Vacuum

NARIT-AI operates in a regulatory grey zone. India currently has no enacted statute specifically governing AI in law enforcement. The Artificial Intelligence (Ethics & Accountability) Bill, 2025 — India's first legislative attempt at AI-specific binding obligations — remains a Private Member's Bill that has not progressed beyond introduction. The IndiaAI Governance Guidelines (November 2025) are voluntary and non-enforceable. This means that if NARIT-AI produces a legally erroneous report that leads to a wrongful prosecution, there is no clear statutory framework to establish liability — whether it rests with the AI developer, the deploying police unit, the individual officer, or the State government.

Issue 1: The Black Box Problem & Explainability Deficit

RAG systems, while more auditable than pure generative AI, still produce outputs through complex probabilistic retrieval processes that are not easily explained in natural language. In a judicial context, this creates a significant problem: if an investigating officer's decisions are guided by NARIT-AI's recommendations, and those decisions are challenged in court, who explains the AI's reasoning? Courts — including the Supreme Court — have consistently held that the burden of proof in NDPS cases is exceptionally high and that any gap in the evidentiary chain can lead to acquittal. An AI-generated report, if not accompanied by a transparent explanation of its analytical basis, could itself become the subject of legal challenge rather than a shield against it.

Issue 2: Algorithmic Bias and Training Data Quality

NARIT-AI is trained on historical High Court and Supreme Court judgments. This raises a fundamental question: if the judiciary itself has historically applied the NDPS Act in ways that disproportionately affected certain socioeconomic or demographic groups, does training on that case law bake those patterns into the AI's recommendations? International research on predictive policing consistently shows that AI systems trained on historical crime data reproduce and amplify existing biases — what is sometimes called a runaway feedback loop. While NARIT-AI's purpose is procedural compliance rather than predictive profiling, the quality and representativeness of its training corpus remain largely opaque to external scrutiny.

Issue 3: Data Privacy and DPDPA Compliance

Each time an officer uploads an FIR into NARIT-AI, sensitive personal data about accused individuals — name, location of seizure, offence details — is processed by an AI system developed in collaboration with a private startup. The Digital Personal Data Protection Act, 2023 (DPDPA), whose Rules were notified in 2025, governs the processing of personal data by state instrumentalities. While the DPDPA provides exemptions for law enforcement investigations, it imposes obligations around data minimization, security safeguards, and breach notification. Whether NARIT-AI's partnership structure with a private company complies with these requirements — particularly regarding the private partner's access to case data — has not been publicly examined.

Issue 4: Deskilling Risk — The Automation Paradox

By providing investigators with ready-made procedural guidance, NARIT-AI may paradoxically reduce the impetus to build genuine legal expertise within the police force. If officers become habituated to outsourcing legal reasoning to an algorithm, the institutional capacity for independent criminal investigation — already stretched — may atrophy further. This is the automation paradox in high-stakes professions: short-term efficiency gains can erode long-term human capability, creating critical vulnerabilities when the technology fails or produces erroneous outputs.

Issue 5: Democratic Deficit — No Legislative Mandate

NARIT-AI was deployed by the Gujarat Police under executive authority, without any specific legislative mandate from the state legislature or parliamentary approval. In a constitutional democracy governed by the rule of law, the deployment of AI systems that influence criminal investigations — with direct consequences for citizens' liberty — arguably requires legislative sanction and parliamentary oversight, not merely executive discretion. The Parliamentary Standing Committee on Communications and Information Technology, in its March 2026 report, recognized AI as a "critical enabler" of policing but also flagged the absence of a comprehensive statutory framework as a significant governance gap.

🔍 Civil Liberties Dimension

The NDPS Act already imposes a reverse burden of proof on the accused in certain circumstances (Section 35 — presumption of culpable mental state). Layering AI-guided prosecution onto this already-harsh regime risks creating an asymmetric justice system where the State's investigative capacity is algorithmically amplified while the accused's ability to challenge procedurally-compliant but substantively unjust prosecutions is structurally weakened. Articles 14 (equality), 20 (protection against arbitrary conviction), and 21 (right to life and liberty) are all potentially implicated.

The central issue is not whether AI can help police — it can, demonstrably. The issue is whether India has the institutional, legal, and ethical infrastructure to deploy AI in criminal justice without creating new injustices in the process of solving old ones.

Implications: Constitutional, Security, Governance & Federalism Dimensions

🔗 Implications — NARIT-AI

Internal Security Implication: Force Multiplication at Scale

If NARIT-AI's impact translates into measurably higher conviction rates in Gujarat's NDPS cases, it could represent a significant upgrade to the State's internal security architecture against the narco-terror nexus. Drug trafficking revenues are known to finance insurgent and terrorist networks operating along India's borders — disrupting the financial pipeline through more effective prosecution is as strategically significant as interdiction. A higher conviction rate also creates a stronger deterrent effect, potentially reducing the appeal of using Gujarat's ports and transport corridors for drug transit operations. The tool's "force multiplier" logic — enabling every officer to perform at specialist level — could address one of the most persistent structural weaknesses of Indian policing: the inverse relationship between investigative quality and geographic remoteness.

Criminal Justice Reform Implication: From Reactive to Proactive Prosecution

NARIT-AI embodies a philosophically significant shift in criminal justice logic: from evidence-gathering followed by legal analysis (the traditional model) to legally-guided evidence-gathering in real time (the AI-mediated model). This has the potential to dramatically reduce the gap between police investigation quality and court-level evidentiary standards — a gap that has historically been exploited by sophisticated defence counsel in NDPS cases. If successful, this model could inform the design of investigation tools for other complex statutes where procedural compliance is critical: PMLA, UAPA, IT Act violations, and cybercrimes.

Federalism & Cooperative Governance Implication

NARIT-AI was developed and deployed by a state police force — a reminder that Police is a State subject under Entry 2, List II of the Seventh Schedule. This means that AI innovation in policing, unlike in defence or intelligence, can proceed at the state level without central government approval. This creates both an opportunity and a risk: the opportunity is that states can innovate rapidly and serve as laboratories of democratic experimentation; the risk is fragmentation, where 28 states develop 28 incompatible AI policing systems with no interoperability, no common standards, and no unified oversight mechanism. The MHA's parallel development of CCTNS 2.0 — a nationally integrated crime tracking system with embedded AI — will need to find a way to interface with state-level tools like NARIT-AI.

🔍 Constitutional Tension

NARIT-AI's design as a "private system" inaccessible to the public or defence raises a constitutional question under Article 21's right to a fair trial. If an AI system shapes the prosecution's strategy — including anticipating and pre-empting defence arguments — and the defence has no access to how that system works, is the adversarial process that underpins India's criminal justice system being compromised? The right to know the case against you is a fundamental component of natural justice; AI-driven prosecution strategy, if opaque, may violate that principle.

Governance & Technology Policy Implication

NARIT-AI's deployment has accelerated a conversation in India about whether AI use in high-stakes governance domains — criminal justice, immigration, welfare, and public safety — requires a dedicated statutory framework with transparency obligations, audit rights, and liability provisions. The Parliamentary Standing Committee's March 2026 report, the IndiaAI Governance Guidelines (November 2025), and the Private Member's AI Ethics & Accountability Bill (December 2025) collectively signal that India's policymakers are converging on the need for binding regulation, even if consensus on the specific architecture has not yet emerged. NARIT-AI is, in this sense, a forcing function for India's AI governance discourse.

🌱 Implication-to-Way-Forward Bridge

The depth of implications — spanning internal security, constitutional rights, federalism, and technology governance — means that NARIT-AI cannot be evaluated on technical merit alone. A holistic assessment requires simultaneously strengthening its effectiveness (better training data, dynamic updates) and its accountability (transparency audits, judicial oversight, legislative mandate). These are not competing goals; they are complementary conditions for the legitimacy of AI in criminal justice.

NARIT-AI's implications radiate outward from a single state tool to encompass India's national security architecture, constitutional principles, federal governance design, and the trajectory of the country's AI governance framework.

Judicial Evolution: The NDPS Case Law Crisis That Made NARIT-AI Inevitable

⚡ Issues — Judicial Dimension

The Pattern: Strong Evidence, Procedural Failure, Acquittal

A systematic reading of India's Supreme Court jurisprudence on NDPS cases from 2020 onwards reveals a stark, consistent pattern: accused persons are being acquitted not because the prosecution failed to establish the presence of contraband, but because investigating officers failed to comply with mandatory procedural requirements during search, seizure, sample collection, and documentation. The Supreme Court has repeatedly and unequivocally stated that procedural safeguards in the NDPS Act are not technicalities — they are essential components of fair and effective justice. This judicial trend is the direct intellectual context that makes NARIT-AI's compliance-engineering approach not just rational, but necessary.

⚖ Landmark Judgment 1

State of Punjab v. Baldev Singh (1999) 6 SCC 172 — Constitution Bench. Held that the right under Section 50 of the NDPS Act (to be searched before a Gazetted Officer or Magistrate) is a fundamental protection and non-compliance with it vitiates the search and seizure entirely, regardless of the quantity of contraband found. Established that Section 50 compliance is non-negotiable — one of the most cited grounds for NDPS acquittals.

⚖ Landmark Judgment 2

Tofan Singh v. State of Tamil Nadu (2020) — (2021) 4 SCC 1 — Three-judge bench (Justice RF Nariman, Justice Navin Sinha; Justice Indira Banerjee dissenting). Held by 2:1 majority that officers investigating NDPS cases are "police officers" within the meaning of Section 25 of the Evidence Act. Consequence: confessional statements recorded under Section 67 of the NDPS Act are inadmissible in evidence and cannot form the basis of conviction. This overruled decades of practice where NCB investigations relied heavily on Section 67 statements. It made it imperative that officers build their cases entirely on search, seizure, and forensic evidence — precisely the procedural compliance domain NARIT-AI addresses.

⚖ Landmark Judgment 3

Yusuf v. State (October 2023) — Supreme Court. Acquitted an accused sentenced to 10 years for possession of commercial-quantity heroin solely on the ground that the prosecution failed to comply with Section 52A (mandatory sampling and documentation of seized drugs). Court held: "the failure of concerned authorities to lead primary evidence vitiates the conviction." Established that procedural violations nullify convictions regardless of the quantity of contraband.

⚖ Landmark Judgment 4

Najmunisha Etc. v. State of Gujarat — 2024 INSC 290 (April 9, 2024) — Two-judge bench. Set aside convictions affirmed by the Gujarat High Court on grounds of non-compliance with mandatory procedural safeguards under Sections 41, 42, and 67 of the NDPS Act. Also reaffirmed the Tofan Singh ruling that Section 67 statements are inadmissible. Directly relevant to Gujarat Police, as this was a Gujarat High Court judgment set aside by the Supreme Court — the very state that subsequently deployed NARIT-AI.

⚖ Landmark Judgment 5 (Most Recent)

Doniyar Vildanov v. State of U.P. — 2026 INSC 95 (February 2026) — Bench: Justice Sanjay Kumar and Justice K. Vinod Chandran. Acquitted a Russian national convicted for 1.9 kg charas recovered at the Indo-Nepal border. Grounds: the mandatory search safeguards under the NDPS Act were "not carried out in their true letter and spirit" — the contraband was searched and detected before the accused was informed of his right to be searched before a Gazetted Officer. The Court observed that the sequence of events narrated by prosecution witnesses contradicted the required procedural order. This 2026 case illustrates that the problem NARIT-AI addresses remains live and urgent.

NDPS Procedural Lapse → Acquittal Matrix — What NARIT-AI Targets
NDPS Section	Requirement	Common Lapse	SC Position
S. 41/42	Written grounds before search/seizure	Grounds recorded after the fact	Mandatory; failure vitiates seizure
S. 50	Inform accused of right to be searched before GO/Magistrate	Not informing accused or inadequate consent recording	Fundamental protection — Baldev Singh 1999
S. 52A	Draw samples, obtain Magistrate certificate for seized drugs	Samples not drawn, improper documentation	Non-compliance vitiates conviction — Yusuf 2023
S. 67	Call for information (not confession)	Using S.67 statements as confessional basis for conviction	Inadmissible — Tofan Singh 2020
Chain of Custody	Unbroken custody of seized drugs from seizure to court	Gaps in documentation, storage failures	Any break can collapse prosecution

✍ Mains Tip — This Panel's Unique Value

No other section of your answer can demonstrate legal depth the way this case law matrix can. In a GS-3 Mains answer on NARIT-AI, citing Tofan Singh (2020) and Najmunisha v. Gujarat (2024) as the specific judicial drivers of the tool's creation elevates your answer from descriptive to analytical. The examiner wants to see that you understand why the tool was needed, not just what it does.

India's Supreme Court did not create the NDPS acquittal crisis — it merely diagnosed and enforced it. NARIT-AI is the police's direct response to that judicial diagnosis: if courts will acquit on procedural grounds, then procedural compliance must be built into the investigation architecture from the very first step.

Initiatives: NARIT-AI Within India's Broader AI-Policing & Governance Ecosystem

🏛 Initiatives — AI in Internal Security & Governance

NARIT-AI in the Context of India's AI-Policing Landscape

NARIT-AI does not exist in isolation — it is the most domain-specific and legally sophisticated of a growing ecosystem of AI tools being deployed across India's internal security apparatus. Understanding this ecosystem is essential for Mains answers that require systemic analysis rather than isolated case description.

India's AI in Internal Security — Comparative Tool Ecosystem (2025–26)
Tool / System	Deploying Agency	Function	Status
NARIT-AI	Gujarat Police	NDPS case analysis, procedural compliance, predictive court strategy	Operational — April 2026
CCTNS 2.0	MHA / NCRB (AI integration)	Crime tracking + predictive policing, criminal profiling, FIR-to-section matching, risk terrain modeling	Under development (2026)
Mule Hunter	I4C (Indian Cyber Crime Coordination Centre)	AI to detect fraudulent bank accounts used in cybercrime (mule accounts)	Operational
SUPACE	Supreme Court of India	AI-assisted case research and legal analysis for judges	Piloted in SC
Adalat AI	State courts	Case management, timeline reduction; deployed in 3,000+ courts across 8 states	Operational (as of Sep 2025)
AMPED FIVE	Delhi Police	AI-enhanced CCTV image clarification and facial matching with government databases	Operational
IVFRT 3.0	MHA (Immigration)	AI + blockchain for intelligent traveller profiling and identity verification	Planned — April 2026
Surakshini	Cyber security apparatus	AI system to prevent spread of harmful online content	Operational

Legislative & Policy Initiatives: The Governance Architecture

The deployment of AI in policing is outpacing the governance frameworks designed to oversee it. Several important policy initiatives are currently active, though none yet provides the statutory accountability framework that the domain requires:

IndiaAI Governance Guidelines (November 2025) — Published by MeitY under the IndiaAI Mission. Voluntary, non-enforceable, principle-based. Establishes an AI Governance Group (AIGG), Technology & Policy Expert Committee (TPEC), and plans a National AI Incidents Database. Proposes a graded liability approach for policing AI.
Artificial Intelligence (Ethics & Accountability) Bill, 2025 — Private Member's Bill introduced in Lok Sabha (December 17, 2025) by an individual MP. Proposes a statutory Ethics Committee, mandatory ethical reviews for high-risk AI in surveillance and law enforcement, documented bias audits (addressing caste, religion, gender, language), and penalties up to ₹5 crore for non-compliance. As of May 2026, has not progressed beyond introduction. Historically, Private Members' Bills rarely become law (last one enacted: 1970).
Parliamentary Standing Committee Report (March 30, 2026) — Report on "Impact of Emergence of AI" tabled in Lok Sabha. Recognized AI as a "critical enabler" in policing. Revealed MHA's plans for CCTNS 2.0 with predictive policing, criminal profiling, and risk scoring capabilities. Flagged absence of comprehensive statutory framework.
Digital Personal Data Protection Act, 2023 (DPDPA) & Rules 2025 — Phased rollout until 2027. Governs AI processing of personal data including law enforcement investigations. Data Fiduciaries processing high-sensitive data face penalties up to ₹250 crore for non-compliance.
IT (Intermediary Guidelines) Amendment Rules, 2026 (February 2026) — Addresses Synthetically Generated Information (SGI), mandatory AI content labelling; does not directly regulate law enforcement AI but signals India's move toward binding AI-specific norms.

Drug Demand Reduction Initiatives — The Prevention Side

NARIT-AI addresses the enforcement and prosecution dimension of India's anti-narcotics strategy. Complementary demand-reduction initiatives provide the broader policy context:

National Action Plan for Drug Demand Reduction (NAPDDR) — Ministry of Social Justice & Empowerment; rehabilitation, awareness, and community-based intervention
Narco Coordination Centre (NARCOD) — Constituted 2016; multi-agency intelligence sharing on drug trafficking
Seizure Information Management System (SIMS) — NCB's online database of drug offences and offenders
National Drug Abuse Survey — MoHFW / AIIMS NDDTC; evidence-based measurement of drug abuse trends
Project Sunrise (2016) — Ministry of Health; tackles HIV cases linked to drug injection in northeast states
Scheme of Financial Assistance to States for Narcotics Control — Centre-to-State funding for anti-narcotics infrastructure

✍ Mains Tip — Nesting NARIT-AI in Ecosystem

Examiners reward answers that contextualise individual initiatives within broader policy ecosystems. Mention CCTNS 2.0, IndiaAI Governance Guidelines, and NAPDDR alongside NARIT-AI to show that you understand India's systemic approach to AI in internal security — not just one tool in isolation. This demonstrates multi-dimensional analytical thinking.

NARIT-AI is the most legally sophisticated point on India's AI-policing curve, but it sits within a rapidly evolving ecosystem where governance frameworks are still catching up with technological deployment — the defining tension of AI in Indian governance in 2026.

Current Affairs: Live Updates — April–May 2026

📊 Current Affairs — DD News · April 10, 2026

NARIT-AI officially launched in Gandhinagar by Gujarat Police on April 10, 2026. Officials confirmed it is "the first kind of AI-based tool being used by any state police or law enforcement agency in India." The tool was launched as part of Gujarat's zero-tolerance approach to drug-related offences, with officials describing it as a "force multiplier" that enables every investigating officer to conduct proper NDPS investigations. The Gujarat High Court's guidelines on AI classified NARIT-AI as a private system, restricting it to verified police users.

📊 Current Affairs — MHA / The Print · March 31, 2026

CCTNS 2.0 to integrate AI for predictive policing and criminal profiling — MHA disclosed to the Parliamentary Standing Committee on Communications and Information Technology that the National Crime Records Bureau (NCRB) has planned multiple AI capabilities for CCTNS 2.0: entity resolution to link cases across states to common suspects, automatic FIR-to-legal-section matching using custom-trained AI-LLM models, Risk Terrain Modeling (RTM) for crime hotspot prediction, and AI-based risk scoring of repeat offenders. The system will link 17,000 police stations across India.

📊 Current Affairs — Parliamentary Standing Committee / MediaNama · March 30, 2026

Parliamentary Standing Committee report recognises AI as "critical enabler" in policing — The report, "Impact of Emergence of Artificial Intelligence and Related Issues," tabled in Lok Sabha, revealed the MHA's planned AI integrations and flagged the absence of a comprehensive statutory framework governing AI use in internal security. The Committee's recognition signals growing parliamentary attention to AI governance in law enforcement contexts.

📊 Current Affairs — Insights IAS / NCB Operation · May 2, 2026

Operation WHITE STRIKE — NCB busts international cocaine trafficking syndicate — A large-scale NCB operation targeted an international cocaine trafficking network operating through India's logistics networks. Coordinated raids in Kalamboli and Bhiwandi warehouses uncovered cocaine hidden in imported machinery cavities with multi-layer packaging. The seizure exceeded India's annual average cocaine recovery. Described as an example of "bottom-to-top intelligence approach" and implementation of India's zero-tolerance policy through intelligence-driven enforcement. Contextually significant alongside NARIT-AI as part of India's multi-pronged anti-narcotics strategy.

📊 Current Affairs — Sanskriti IAS / Vajiram & Ravi · April 3, 2026

MHA expanding AI across internal security domains — The Ministry of Home Affairs confirmed it is significantly expanding AI deployment across predictive policing, cybercrime detection, and financial fraud prevention. The IVFRT 3.0 system (Immigration, Visa, Foreigners Registration and Tracking), planned for April 2026 rollout, integrates AI and blockchain for intelligent traveller profiling and record authentication — extending AI-enabled security from domestic criminal investigation to border and immigration management.

📊 Current Affairs — Supreme Court / BharatLaw.ai · February 2026

Doniyar Vildanov v. State of U.P. — 2026 INSC 95 — The Supreme Court acquitted a Russian national convicted for charas recovery at the Indo-Nepal border, citing failure to follow mandatory NDPS search and seizure safeguards. The Bench noted that the "mandatory stipulation for search and seizure as per the NDPS Act was not carried out in its true letter and spirit." This February 2026 acquittal — just two months before NARIT-AI's April 2026 launch — powerfully contextualises the continuing urgency of the procedural compliance problem the tool addresses.

✍ Mains Tip — Current Affairs Integration

The Doniyar Vildanov acquittal (February 2026) followed by NARIT-AI's launch (April 2026) creates a cause-and-effect narrative that is powerful in Mains answers. Mention both: "As recently as February 2026, the Supreme Court acquitted an accused despite contraband recovery, citing procedural non-compliance — the precise problem NARIT-AI was launched to solve two months later." This temporal precision demonstrates current affairs awareness and analytical linkage simultaneously.

India's April–May 2026 landscape shows a clear convergence: state-level AI tools (NARIT-AI), central AI infrastructure (CCTNS 2.0), active anti-narcotics operations (WHITE STRIKE), and live judicial pressure (Doniyar Vildanov) are all pulling in the same direction — toward a technology-enabled, legally rigorous, and conviction-focused approach to narcotics enforcement.

Quick Revision & Answer Framework: NARIT-AI for Mains Mastery

💡 Innovation & Way Forward — NARIT-AI

⚡ Rapid Recall — NARIT-AI (Internal Security · Mains GS-3)

Full Form: Narcotics Analysis & RAG-based Investigation Tool — Artificial Intelligence
Launched: 10 April 2026, Gandhinagar · by Gujarat Police · First such state police AI tool in India
Developed by: Western Railway Police (Vadodara Division) + Gradiante Creative Services (Mumbai AI startup)
Core Technology: Retrieval Augmented Generation (RAG) — closed database, not open internet — eliminates hallucination risk
Training corpus: NDPS Act 1985 · BNS · BNSS · BSA · SC/HC landmark judgments · Government circulars
Trigger crisis: Gujarat NDPS conviction rate fell to 25% by 2022 — primarily due to procedural lapses by investigating officers
Key judicial backdrop: Baldev Singh (S.50) · Tofan Singh 2020 (S.67 confessions inadmissible) · Yusuf 2023 (S.52A) · Najmunisha v. Gujarat 2024 · Doniyar Vildanov 2026 INSC 95
Constitutional anchor: Article 47 (DPSP — prohibition of harmful drugs) · Articles 14, 20, 21 (civil liberty tensions)
Key features: FIR analysis → case strength/weakness report → evidence checklist → predictive court strategy → draft chargesheet
Security design: Classified as private AI system (Gujarat HC guidelines) · Double-layer encryption · Accessible only to verified officers
Governance gap: No enacted AI-specific statute · AI Ethics Bill 2025 = Private Member's Bill (not law) · IndiaAI Guidelines = voluntary only
Ecosystem context: CCTNS 2.0 (MHA/NCRB) · Mule Hunter (I4C) · SUPACE (SC) · Adalat AI · IVFRT 3.0 · Operation WHITE STRIKE (NCB May 2026)

🎯 NARIT-AI answer hook: "When a 25% conviction rate reveals that India's war on drugs is being lost not in the field but in the courtroom, the answer is not more police — it is smarter compliance."

· MaargX UPSC · Curated for Civil Services Preparation ·

📝 Mains Answer Framework — NARIT-AI (150 / 250 words) · 5I Approach

📖 Introduction

Open with the conviction rate crisis: Gujarat's NDPS conviction rate of 25% (2022), driven by procedural lapses, created an institutional imperative. In April 2026, Gujarat Police launched NARIT-AI — India's first state police AI tool for narcotics investigation — built on Retrieval Augmented Generation technology. Its launch, just two months after the Supreme Court's Doniyar Vildanov acquittal (February 2026), underscores the systemic urgency it addresses.

⚡ Issues

Three core issues: (1) The procedural lapse crisis — SC rulings in Tofan Singh (2020), Yusuf (2023), and Najmunisha (2024) established that non-compliance with Sections 41, 42, 50, 52A, and 67 of the NDPS Act vitiates convictions regardless of evidence strength. (2) The accountability vacuum — no enacted statute governs AI in law enforcement; the AI Ethics Bill 2025 remains a Private Member's Bill. (3) Civil liberties tensions — NARIT-AI's closed-system design raises Article 21 concerns about fair trial rights in an adversarial legal system.

🔗 Implications

Positive: Force multiplication enabling any-rank officers to investigate at specialist level; potential to disrupt narco-terror financing by raising conviction rates; model for other complex statutes (PMLA, UAPA). Concerning: Risk of algorithmic bias embedded in historical case law; deskilling of investigative capacity; federalism fragmentation as 28 states may develop incompatible AI systems without national standards.

🏛 Initiatives

NARIT-AI sits within a growing ecosystem: CCTNS 2.0 (MHA/NCRB, under development — predictive policing + criminal profiling); Mule Hunter (I4C); SUPACE and Adalat AI (judiciary); IVFRT 3.0 (immigration AI + blockchain). Governance initiatives include IndiaAI Governance Guidelines (November 2025, voluntary), AI Ethics & Accountability Bill 2025 (pending), and DPDPA Rules 2025. Parliamentary Standing Committee (March 2026) recognises AI as a "critical enabler" in policing.

💡 Innovation

Way forward: (1) Enact binding legislation — AI in law enforcement must have a statutory mandate, not just executive deployment; include transparency obligations and bias audit requirements. (2) Human-in-the-loop mandate — AI recommendations must remain advisory; no prosecution decision should be AI-automated. (3) National interoperability standard — MHA should set common technical and ethical standards for state police AI tools to prevent fragmentation. (4) Independent audit mechanism — third-party evaluation of NARIT-AI's accuracy, bias, and legal compliance, accessible to the judiciary. Ultimately, NARIT-AI's legitimacy rests on aligning Article 47's anti-narcotics mandate with Articles 14 and 21's due process guarantees — technology cannot substitute for constitutional balance.

🌱 Way Forward — Detailed Reform Roadmap

Statutory mandate for AI in policing — Parliament should enact an AI in Law Enforcement (Transparency & Accountability) Act, modelled on the UK's approach to algorithmic policing oversight, requiring pre-deployment review for any AI system influencing criminal investigations
Human-in-the-loop as non-negotiable — NARIT-AI outputs must be treated as advisory reports, not binding determinations; investigating officers must certify independent application of legal judgment
National AI Policing Standards Body — MHA should establish an inter-state technical committee to define interoperability, data sharing, and ethical standards for all state-level policing AI tools, preventing the balkanization of India's law enforcement AI ecosystem
Open-source audit of training data — NARIT-AI's judgment corpus should be subject to periodic independent review for representation bias, outdated precedents, and coverage gaps in new legal frameworks (BNS/BNSS/BSA)
Defence access mechanism — Consistent with Article 21's right to a fair trial, a mechanism should be developed to allow defence counsel to seek judicial review of AI-generated reports used in investigations, ensuring adversarial integrity of criminal proceedings
Expand to other complex statutes — NARIT-AI's success, if verified, provides a replicable model for PMLA, UAPA, IPC offences involving forensic evidence, and cybercrime investigations under the IT Act

The measure of NARIT-AI's success is not whether conviction rates rise — it is whether they rise because of genuinely stronger prosecutions, not because an algorithm has made it harder for the accused to exercise their constitutional rights.