AI in Defence (UPSC Science & Tech)
UPSC guide to AI in defence: LAWS, DAIC/DAIPA, India's AI defence roadmap, deepfakes, cyber warfare, and 2024-26 developments.
Modern militaries are embedding Artificial Intelligence (AI), machine learning, autonomy and data analytics into every element of combat — from intelligence collection to command and control, autonomous platforms, cyber operations and information warfare. India has been steadily building a defence AI roadmap since 2018, with institutional mechanisms like the Defence AI Council (DAIC) and Defence AI Project Agency (DAIPA). For UPSC, this is a key GS Paper III (Science & Technology, Internal Security, Defence) topic with GS Paper IV (ethics) overlap.
Why AI in defence?
- Speed — AI processes vast data faster than humans.
- Precision — machine vision for targeting and threat identification.
- Persistence — 24×7 surveillance with no fatigue.
- Scale — swarms of drones and autonomous vehicles multiply force.
- Decision support — commanders get real-time, option-weighted recommendations.
- Asymmetric edge — offsets numerical disadvantages via tech superiority.
Areas of defence AI
| Area | Application |
|---|---|
| Intelligence, Surveillance, Reconnaissance (ISR) | Satellite & UAV imagery analytics; SIGINT |
| Autonomous & Unmanned Systems | Drones, UGVs, UUVs |
| Command & Control (C2) | Battle management, COP fusion |
| Cybersecurity | AI-driven threat detection and response |
| Logistics & Maintenance | Predictive maintenance, supply chain optimisation |
| Training & Simulation | AI-generated scenarios, VR/XR |
| Electronic Warfare | Cognitive jamming and counter-jamming |
| Information Warfare | Deepfake detection, counter-propaganda |
| Lethal Autonomous Weapon Systems (LAWS) | Autonomy in target engagement |
| Internet of Battle Things (IoBT) | Sensor-rich kits; smart helmets |
| Health & Medical | Triage, telemedicine, wearables |
Global AI defence landscape
- USA — Project Maven (ISR fusion), JADC2 (Joint All-Domain Command and Control), Replicator Initiative (thousands of autonomous systems by 2025), AUKUS pillar 2.
- China — PLA's intelligentised warfare doctrine; civil-military fusion; record AI patents.
- Russia — autonomous platforms, electronic warfare in Ukraine theatre.
- Israel — IDF's Lavender and Gospel AI target-nomination systems (Gaza, 2024).
- Turkey, Iran — loitering-munition drones widely exported.
- NATO — DIANA, NATO AI Strategy 2021 & Generative AI addendum 2024.
India's AI defence architecture
| Body | Function |
|---|---|
| Defence AI Council (DAIC) | Apex policy, chaired by Raksha Mantri |
| Defence AI Project Agency (DAIPA) | Project identification, budgeting |
| AI Task Force (Baba Kalyani committee 2018) | Foundational strategy |
| iDEX / iDEX-Prime / ADITI | Defence-tech startups |
| DRDO's Centre for AI & Robotics (CAIR), Bengaluru | R&D on autonomy, swarms |
| Army Design Bureau, Indian Navy Innovation & Tech Dev. Scheme (INEFAD), IAF Innovation Cell | Service-level innovation |
| Defence Cyber Agency (DCA) | Cyber warfare domain |
| Defence Space Agency (DSA) | Space domain |
| Integrated Defence Staff, CDS | Integration & doctrine |
Key programmes
- ~75 AI products roadmap (2022) released by Raksha Mantri; covers surveillance, cyber, autonomous systems.
- Each service earmarked Rs 100 crore/year for AI adoption (operational from 2020 onwards).
- iDEX DISC 12 and ADITI Scheme (Acing Development of Innovative Technologies with iDEX, 2024) — Rs 750 crore for critical & emerging tech startups.
- DRDO Technology Development Fund — AI projects approved.
- Swathi, Ashwin, Pinaka — integrated AI-enhanced fire-control systems.
- Indigenous autonomous platforms — DRDO's AURA, SWiFT, Ghatak UCAVs; Army's remotely-operated vehicles; Navy's underwater drones.
Lethal Autonomous Weapon Systems (LAWS)
LAWS are systems that, once activated, can select and engage targets without human intervention. They raise foundational legal and ethical questions under International Humanitarian Law (IHL):
- Principle of distinction — can machines distinguish combatants from civilians?
- Principle of proportionality — can an algorithm weigh military necessity vs collateral harm?
- Accountability gap — who is responsible — operator, commander, programmer, manufacturer?
- Arms race risk — unregulated deployment lowers threshold for conflict.
Governance efforts
- UN CCW (Convention on Certain Conventional Weapons) Group of Governmental Experts (GGE on LAWS) — talks since 2014; no binding treaty yet.
- "Stop Killer Robots" campaign — civil society.
- Political Declaration on Responsible Military Use of AI and Autonomy (REAIM Summit, The Hague 2023 and Seoul 2024) — ~60 signatories including India.
- UNGA First Committee Resolution (Dec 2023) — calls for a report on LAWS.
Deepfakes and information warfare
- Deepfakes — synthetic media generated via GANs and diffusion models.
- Used in disinformation, blackmail, election interference.
- India's 2024 Lok Sabha election saw 50+ manipulated political videos; ECI, MeitY, I&B coordinated take-downs.
- Legal response: IT Act Section 66D (cheating by personation), IT Rules 2021, proposed Digital India Act, MeitY Advisory March 2024, BNS Section 318 (cheating).
- Detection tools — emerging — watermarking standards by C2PA; DeepFake-o-meter.
Cyber warfare and AI
AI both enables and defends against cyber attacks.
- Offensive — automated phishing, polymorphic malware, AI-guided intrusion.
- Defensive — anomaly detection, threat hunting, auto-response (SOAR).
- India's Defence Cyber Agency (DCA) coordinates offensive-defensive cyber ops.
- CERT-In, NCIIPC, I4C handle civilian cyber.
Internet of Battle Things (IoBT)
A network of sensors, wearables, weapons and vehicles linked to a common tactical cloud.
- Smart helmets — augmented reality, biometric monitoring.
- Battlefield edge clouds for near-sensor AI.
- 5G tactical networks — private slices for secure comms.
- Integrated Battlefield Management Systems (BMS).
Challenges
| Challenge | Detail |
|---|---|
| Data quality | Military data is siloed and classified; hard to train AI |
| Talent | Scarce AI engineers in cleared environments |
| Explainability | Black-box AI unacceptable in lethal decisions |
| Cybersecurity | AI models themselves attackable (model poisoning, adversarial inputs) |
| Dual-use | Imported AI components could embed backdoors |
| International law | LAWS normative vacuum |
| Ethical risk | Deepfake escalation, bias, dehumanisation of war |
| Human-machine teaming | Trust calibration is hard |
India's strategy — principles
From the 2018 AI Task Force and subsequent MoD statements:
- Human-in-the-loop in lethal decisions.
- Indigenous development for sovereign tech.
- Private-public ecosystem — iDEX, ADITI.
- Civil-military fusion via IndiaAI Mission compute.
- Ethics and IHL compliance.
- International engagement — REAIM, UN GGE, Quad critical tech.
Latest developments (2024-26)
- ADITI Scheme (Mar 2024) — Rs 750 crore for deep-tech defence.
- REAIM Seoul Declaration (Sept 2024) — India signatory; responsible military AI principles.
- India's AI Summit (Dec 2024) previewed AI Action Summit Paris (Feb 2025); IMoD pitched human-in-the-loop.
- Multi-domain Command & Control Pilot — Theatre Commands integration experiments.
- Ghatak UCAV — technology demonstrator flight tests progressed.
- Indian Army Drone Battalion for each Infantry Battalion — raised (2024-25).
- Advanced Towed Artillery Gun System (ATAGS) with AI-based fire control inducted (2024).
- Indo-US iCET — defence AI critical technology track.
- India Semiconductor Mission — rad-hard, secure AI chips for defence platforms.
- DeepTech Policy 2024 — defence deep-tech start-ups prioritised.
- National Quantum Mission (2024) — quantum sensing for navigation warfare.
- Chandrayaan-4 plan and Gaganyaan progress — spin-offs in autonomous guidance and robotics.
- Deepfake rules (proposed 2025) — mandate watermarks for AI-generated political content.
UPSC Relevance
GS Paper III — Science & Technology / Internal Security / Defence
- AI in warfare, LAWS, cyber, deepfakes.
GS Paper II — International Relations
- UN CCW GGE, REAIM, Quad critical tech, iCET.
GS Paper IV — Ethics
- Autonomy in lethality, accountability, civilian harm.
Essay
- "When the machines learn to fight, who takes responsibility?"
Prelims pointers — DAIC, DAIPA, CAIR DRDO, iDEX/ADITI, LAWS, UN CCW GGE, REAIM Seoul Declaration, Section 66D IT Act, IT Rules 2021.
Interview probes — India's LAWS stance; human-in-the-loop in hot conflicts; deepfake governance; civil-military AI fusion; iDEX effectiveness.
AI will shape the next generation of warfare as profoundly as gunpowder, aircraft or nuclear weapons. India's advantage lies not in outspending adversaries — but in sovereign talent, ethical anchors and integrated civil-military innovation. That is the strategic opportunity.