AI in Surveillance and OSINT: Privacy Law and Proportionality in India

AI makes it possible to watch, match and profile people at a scale that manual OSINT never could, through face matching, cross-platform account linking and automated monitoring of posts.

Mar 06, 2025 - 12:29
Updated: 
102.4k

AI in Cybersecurity in Cybersecurity & Ethical Hacking · Explainer

Security Tools, Authorized Labs & CTFs

AI in Surveillance and OSINT: Privacy Law and Proportionality in India

Quick answer: AI makes it possible to watch, match and profile people at a scale that manual OSINT never could, through face matching, cross-platform account linking and automated monitoring of posts. In India, the Supreme Court’s 2017 Puttaswamy judgment recognised privacy as a fundamental right and set a proportionality test: any intrusion needs a legal basis, a legitimate aim, a measure no wider than necessary and safeguards against abuse. The DPDP Act 2023 adds duties for anyone processing personal data. Organisations using AI for OSINT or monitoring should define a narrow purpose, minimise data, document their legal basis and keep a human accountable.

Key takeaways

  • AI turns scattered public data into detailed profiles, which raises privacy risk.
  • Puttaswamy made privacy a fundamental right with a proportionality test.
  • The DPDP Act 2023 applies to digital personal data, with some exemptions.
  • Collecting public data at scale can still be unlawful or harmful.
  • Purpose limits, minimisation and oversight keep OSINT work defensible.

Table of Contents

How Is AI Changing Surveillance and OSINT?

Earlier OSINT relied on analysts searching sources by hand. AI now automates monitoring of many platforms, matches faces and voices, links accounts by writing style and builds timelines from scattered posts. Our article on AI in reconnaissance and data gathering covers these capabilities. Scale is the key change: combining many harmless facts can reveal where someone lives, works and worships.

Where Is the Line Between OSINT and Surveillance?

FactorBounded OSINTSurveillance
PurposeSpecific question, such as a phishing domainOngoing watching of people
TargetInfrastructure or a defined caseIndividuals or groups broadly
DurationEnds when the question is answeredContinuous
Data keptOnly what the case needsEverything collected
Legal basisDocumented and narrowOften unclear

Lawful interception and monitoring by the state is governed by statute, including provisions of the IT Act 2000, and is outside what private organisations may do.

What Did the Puttaswamy Judgment Decide?

On 24 August 2017 a nine-judge bench of the Supreme Court held in Justice K.S. Puttaswamy (Retd.) v. Union of India that privacy is a fundamental right under the Constitution, as listed in the court’s landmark judgment summaries. The judgment discussed informational privacy and the need for a data protection law. Restrictions on privacy are tested for proportionality, commonly described in four parts:

  1. Legality: the measure has a basis in law.
  2. Legitimate aim: it pursues a proper purpose.
  3. Necessity and proportionality: it is no wider than needed.
  4. Safeguards: there are procedures against abuse.

How Does the DPDP Act Apply to AI-Driven OSINT?

The DPDP Act 2023 governs processing of digital personal data. It sets duties on data fiduciaries such as purpose limitation, security safeguards and breach notification, and gives individuals rights over their data. The Act includes exemptions, including powers for the government to exempt certain state bodies and processing. Its treatment of personal data that a person has made public is a detail that needs careful legal reading for each use case. Organisations should not assume that scraping public profiles is automatically allowed.

What Are the Main Privacy Risks?

RiskExample
ProfilingCombining posts to infer religion, health or politics
MisidentificationFace match errors pointing to the wrong person
Chilling effectPeople stop speaking freely if they feel watched
Function creepData collected for security reused for other aims
Data breachesLarge OSINT stores become targets themselves

The UN human rights office tracks these concerns in its work on privacy in the digital age.

How Can Organisations Apply a Proportionality Check?

  • Write down the purpose and legal basis before collection starts.
  • Target infrastructure and threats, not individuals, wherever possible.
  • Collect the minimum, set retention limits and delete on schedule.
  • Run a privacy risk review using the NIST Privacy Framework and the NIST AI RMF.
  • Require human review before any action affecting a person.
  • Follow investigation standards such as the Berkeley Protocol.

Never access accounts or systems without permission; that can breach the IT Act, Sections 43 and 66. See also our post on ethical challenges of AI-powered OSINT.

What Should Practitioners Expect Next?

Expect closer scrutiny of AI profiling, more questions from clients about data handling and more demand for staff who understand both OSINT and privacy law. Record your reasoning now so your work stands up later. To learn responsible OSINT methods, ask about our OSINT and privacy training.

References

Frequently Asked Questions

AI automates monitoring of online sources, matches faces and voices, links accounts across platforms and builds profiles or timelines from scattered data.

In August 2017 a nine-judge Supreme Court bench held that privacy is a fundamental right under the Indian Constitution, with restrictions tested for proportionality.

A privacy restriction must have a legal basis, a legitimate aim, be no wider than necessary and include safeguards against abuse.

Not automatically. The DPDP Act 2023, platform terms and the purpose of collection all matter, so take legal advice before large-scale collection.

It can, because OSINT often processes digital personal data. Duties such as purpose limitation and security apply unless a specific exemption covers the processing.

Bounded OSINT answers a specific question and keeps only needed data, while surveillance involves ongoing watching of people, often without a clear limit.

Yes. Face matching makes errors, especially with poor images, so a match should be treated as a lead that needs independent confirmation.

Document purpose and legal basis, target threats rather than people, minimise and delete data on schedule, review privacy risk and keep human oversight.

What's Your Reaction?

Like Like 0
Dislike Dislike 0
Love Love 0
Funny Funny 0
Wow Wow 0
Sad Sad 0
Angry Angry 0
Vaishnavi

Vaishnavi is a skilled tech professional at the Ethical Hacking Training Institute in Pune, responsible for managing and optimizing the technical infrastructure that supports advanced cybersecurity education. With deep expertise in network security, backend operations, and system performance, she ensures that practical labs, online modules, and assessments run smoothly and securely. Her behind-the-scenes contributions play a vital role in delivering a seamless and secure learning experience for aspiring ethical hackers.