Artificial Intelligence & Security

Introduction:

• The world order is appearing to be transiting from the Age of Information to the Age of Artificial Intelligence (AI). The mandate of the Age of Information was that one had to be well informed for achieving success in any field since knowledge-based decision-making was the key to that success. In the given context, the Information can be defined as data or a fact that helps to make a difference 'between a decision and a guess'.
• As Intelligence is the information about 'what lies ahead' it can be considered as the foundation of effective 'action' in any sphere, be it security. The important thing is that today there is a newfound significance attached to the connectivity between Intelligence, Decision and Response.

In this piece, we will try to explore how AI technology would be able to contribute to India’s national security.

Understanding Artificial Intelligence (AI):  

• The birth of Artificial Intelligence (AI) can be traced back to the 1950s in an American computer scientist Alan Turing’s essay on ‘Computing machinery and intelligence’. Later on, the term AI was used as a title of a conference held at Dartmouth College in 1955. In this historic conference, John McCarthy (computer scientist), imagining a great collaborative effort, brought together top researchers from various fields for an open-ended discussion on artificial intelligence, the term which he coined at the very event Since then, the development in AI started.
• The definition of Artificial Intelligence is the ability of the computer system to perform tasks that normally required human intelligence. Automation machine learning and deep learning are significant areas of AI.
• AI also refers to the ability of machines to perform cognitive tasks like thinking, perceiving, learning, problem-solving and decision-making as a constellation of technologies that enable machines to act with higher levels of intelligence.

Use of Artificial Intelligence in Security:

• Security has two components – accessing information about the secretive adversary and choosing the course of action to neutralize the latter. Artificial Intelligence generally bolsters the nature of 'response' in the sphere of security. AI can help to sharpen the instruments of 'action'. Technology makes a process 'smart' and Artificial Intelligence is all about making technology 'smarter'. The principal utility of AI there is in the collation and analysis of information available. National Security calls for accessing information about the well-guarded secrets of the enemy about which not much would be found in the examination of open information.
• In the age of the Internet, no one can deny that there is always a scope for 'footprints' of the adversary being left behind in that medium and that is why a scan of various communication systems receives priority in the security set-up. The difficult task of collecting human intelligence through secret means still remains the principal objective of Intelligence agencies. It is also true that the human part of intelligence collection can never be entirely substituted by a technology-based endeavour like Artificial Intelligence for achieving that purpose.

The penetration of AI into our systems has recently exploded due to newer developments like unlimited access to computing power, a huge fall in the cost of storing data and the explosion of digitized and structured data. Following are a few important areas where AI can be deployed:

• Processing large amounts of data and recognizing patterns: Artificial Intelligence excels at finding insights and patterns in large data sets that humans just can't see. AI is capable of unifying data across different sources and its algorithms are designed to produce findings in real-time.
• Unlike passive machines that can produce predetermined readings, AI-assisted systems can process data coming in from digital information, satellite imagery, visual information, text or unstructured data and produce an instant signal. The success of AI, however, depends ultimately on the perfection of human input behind all algorithms.
• The big data analytics associated with Artificial Intelligence has affected analysis and assessment functions to a great extent in various fields but at the same time, the area of national security cannot do without the 'human oversight and control' being placed on top. AI can be deployed through the massive data and video captured by surveillance primarily for alerting 'human analysts' about 'patterns' or abnormal and suspicious activity.
• The era of Proxy Warfare: With the advent of 'proxy wars', the ethical dimensions of combat had taken a back seat adding to the unpredictability of the enemy's moves and this is what makes both Intelligence collection and strategy of action an arduous challenge there.
• In a proxy or virtual warfare, the initiator of the offensive had an advantage but AI helps the efficacy of the counter-offensive by identifying the sources behind the attack amidst haze and showing the way for a 'punitive' response to create deterrence.

The use of drones and quadcopter for surveillance and the surreptitious dropping of arms and explosives in Kashmir illustrates the new dimensions of a proxy war that made the task of security so much more challenging.

• The doctrine of 'deterrence': In regard to nuclear arms, the doctrine of 'deterrence' keeps a country from provoking a war. However, secret attacks seem to be the order of the day and these have added to the importance of AI-assisted systems of Intelligence collection and collation as also of counter-measures.

Potential Challenges in the Adoption of AI in National Security:

• Lack of Critical Infrastructure: As AI runs on complex algorithms on loads of data, it is essential to have robust hardware and enable data banks within the country. To process and transmit this time-sensitive data from the sensors and data banks to the command centres India needs a secure and high-speed network.
• Development of Ethical Standards: use of AI in defence would raise a large number of ethical questions like- who holds the accountability in case AI does not perform as predicted? How can AI be integrated with current protocols followed in the forces? The development of these ethical standards is a prerequisite for the adoption of AI for National Security.
• Increase Vulnerability to Cyber-Attacks: The proliferation of AI systems will increase the number of ‘hack-able things’ including systems that kinetic energy (moving vehicles) which may in turn allow exploitive actions to induce lethal effects.
• Theft vulnerability: AI systems are particularly vulnerable to theft by virtue of being almost entirely software-based.
• Limited Role of the Private sector in Defence
• Technology cannot be completely controlled

ARTIFICIAL INTELLIGENCE IN INDIAN ARMED FORCES:

A drastic rise in violence by terrorist groups globally has put an intense focus on border security to combat recent national security issues. Indian armed forces have to manage its 15000 KMs land border and 7516 KMs marine borders; it also has to surveillance vast airspace. The smart and adequate use of AI in armed forces would enhance the capabilities and effectiveness of all forces.

• Potent Delivery System: Logistic management is the backbone of force structure. The AI can be utilized for logistic maintenance in forces for supply chain management.UAVs have emerged as the most potent delivery platforms.
• It has the potential to reduce the number of supplies stored at the frontline and increase the speed of resupply. Drones and unmanned helicopters can be utilized to supply regular accessories like food, clothes, medicines, and ammunition to the forward posts in the border region. In India’s context, it can be used to supply high-altitude terrain in the Siachen glaciers.
  
  

  India’s force structure is in a transition phase. It has decided to establish five theatre commands; Northern command, Western command, Peninsular command, Maritime command, and full-fledged unified Air defence command. The logistic management should be mobile enough to fulfil the requirement of all three wings of the armed forces under one single command. Thus, in such scenarios, the use of AI would enhance synergy in logistic management.

• Reducing the embedded risks during operations: The development of uninhabited aerial vehicles (UAVs) raises the possibility that states will be able to conduct military operations in a more efficient and less risky fashion than was the case when aircraft were piloted by humans. UAVs and drones can be used for surveillance and reconnaissance.
• Using advanced, autonomously navigating air and land vehicles, aid missions can be conducted at greater speed. It will also support troops in future military operations, saving lives by reducing risk and increasing the pace of operations. The small drones can be used in tactical operations by forces. It would help to provide real-time intelligence for tactical decision-making by commanders. The firepower of UAVs is lethal enough to attack enemy tanks and artillery systems.
• Situational Awareness: Situation awareness is very crucial in securing borders in times of stand off’s and engagement of forces. situational awareness relies heavily on Intelligence, Surveillance, and Reconnaissance (ISR) operations.
• Enhance border management architecture: The use of radars and sensors to monitor India’s land borders would eventually enhance border management architecture. It would help to monitor and contain infiltration on border areas. The forces can also use ‘face recognition’ system platforms to detect the enemy in surveillance by Drones and UAVs.
• Integration into close combat battles: The use of robots would be integrated into the close combat battle; robots would be able to enter into the target area and attack on adversaries. Its capacities are more lethal than humans. These technologies can effectively be used in counter-insurgency and anti-terror operations by armed forces and paramilitary forces.
• Information warfare, Algorithmic warfare: These are emerging tactics of modern-day warfare, for example, ‘Grey zone warfare’.
• Grey zone warfare is a mixture of psychological operations, media warfare, information warfare etc. The intensity of grey zone warfare is so effective that it can internally dis-stabilize adversaries. It is majorly used in urban warfare.
• Pakistan is frequently using grey zone warfare tactics against India with regard to Jammu and Kashmir. Thus, Indian policymakers have to think about looking into social media, information and mass-media affairs from a military and security perspective.
• Intent-Based Network Security (IBNS): New AI applications are emerging in Intent-Based Network Security (IBNS) on AI platforms for cyber defence or immune computer systems which have the ability to self-adapt, which helps in countering complex cyber threats, malware reverse engineering and projection to enhance cyber situation awareness. Crowdstrike and DarkTrace software which used AI effectively to monitor cyber threats, leading to huge valuations for the parent companies. The CIA, MIS, NSA and GCHQ were among its clients.
• ‘Cold start’ doctrine: The Integration of AI in the battle tanks, as well as communication between tanks and drones, would provide the upper hand to the Armoured corps. Another point the Indian Army will have to consider is the use of drones against tanks, artillery and manned force. Thus, it is likely that Pakistan can use drones and UAVs to counter India’s cold start doctrine.
  
  

  'Cold Start' suggests the wish to avoid a full-scale 'hot' war. The idea for the Cold Start was fuelled by Operation Parakram, launched after the terror attack on Parliament in December 2001. It means Indian forces making swift and hard inroads into Pakistan. Such strikes will be limited in scope so as not to give any reason to Pakistan to launch a full-scale retaliation. Surprise is a key element of the Cold Start doctrine. In traditional offences, the mobilisation of troops takes a lot of time. By then, the enemy country can not only prepare for a response but also activate international diplomatic channels to prevent an Indian attack.

• Integration of AI technology in the naval platform: The use of unmanned submarines would be an integral part of naval warfare. In 2020, China has successfully tested 56 ‘swarm sharks’, which can be integrated into the PLA Navy. The integration of AI technology in naval platforms would boost the Indian Navy’s capabilities in the Indian Ocean region.
• The AI technologies such as unmanned patrol vessels, unmanned helicopters, drones, and UAVs can be used for surveillance and reconnaissance. The unmanned submarines or unmanned undersea vehicles (UUVs), equipped with ammunitions and short-range torpedoes would be a major deterrence to the adversaries. It would be able to provide the upper hand for sea control to the Indian Navy.
• Enhance the Airforce’s capability to acquire the target: The integration of AI would enhance the Airforce’s capability to acquire the target, and to avoid the enemy’s air defence radars, etc. The AI technology also suggests the use of weapons to fighter pilots. In an adverse situation or dog fight, it provides an upper hand to the pilot. The US in its F-35’s Automatic Logistics Information System (ALIS) uses real-time data from the aircraft’s engine and other onboard systems to predict the next maintenance date and the issues that need to be addressed.
• Air Defence and Nuclear Command and Control System: India also has a ballistic missile defence program. Several missile defence systems and missiles are integrated into the services. India can enhance their capabilities by inducting AI applications into the missile system or in command-and-control platforms to make accurate decisions based on processed information.
• Automation of weaponry system: AI technology that can be integrated into platforms or in particular equipment, thereafter equipment can perform on its own. The AI technology can be used in rockets, missiles, radars, fighter aircraft, and naval platforms like submersible ships, aircraft carriers, frigates and destroyers.
• Improved decision-making: The integration of AI in C4I2SR (command, control, computers, communication, intelligence, information, surveillance, reconnaissance) effectively enhances mobility in decision-making all the time. Image recognition, text analysis, self-driving cars, and game-playing are examples of the recent development in AI which are driving military applications.
• Cyber security: Cyber-attacks have been occurring with increasing frequency. For Example, the leak of personal information of 3.2 million debit cards in 2016 and the Data Theft at Zomato (2017), WannaCry Ransomware (2017), PETYA Ransomware (2017) etc. AI is used in cyber security to quickly analyse millions of events and identify many different types of threats – from malware exploiting zero-day vulnerabilities to identifying risky behaviour that might lead to a phishing attack or download of malicious code.

Steps Taken by India to Leverage AI for Military Applications:

• The Indian defence establishment, on its part, now has a Defence Artificial Intelligence Council (DAIC) with the defence minister as chairman and the three service chiefs, the defence secretary, and the secretary of defence production as members. Defence minister Rajnath Singh had earlier declared that “25 defence-specific AI products” will be developed by 2024.
• A Defence AI Project Agency (DAIPA) has also been created under the secretary (defence production), with Rs 100 crore earmarked annually for AI-enabled projects.

• Wargame Research and Development Centre (WARDEC): Recently, the Army Training Command has signed an MoU with Rashtriya Raksha University (RRU) to develop a Wargame Research and Development Centre (WARDEC) in New Delhi.
• RRU is an Institute of National Importance located in Gandhinagar and established by the Government of Gujarat. The responsibility to prepare the laboratory lies with Tech Mahindra and the centre will be used to develop new war strategies, practise tactics, and assess combat formations in order to improve warfare skills.
• Defence India Startup Challenge: In a recent move, Defence Minister Rajnath Singh kickstarted the sixth edition of the Defence India Startup Challenge under the Innovations for Defence Excellence (iDEX) programme.
• The aim is to fund startups that address AI, sophisticated imaging, sensor systems, big data analytics, autonomous unmanned systems, and secure communication systems, among other technologies for the defence forces. This program is designed to help projects with budgets ranging from Rs 1.5 crore to Rs 10 crore. The iDEX programme is providing our science and technology professionals with an opportunity to understand futuristic technologies such as AI, augmented reality, block-chain and space technologies.
• India-US Defense Artificial Intelligence Dialogue: During the recent fourth U.S.-India 2+2 Ministerial Dialogue between India and US Defence and External Affairs Minister, there was a call for increased collaboration. A further agreement was made to establish the Defence Artificial Intelligence Dialogue and to increase collaborative cyber training and exercises.

Legal and Regulatory Approaches for Managing AI Systems:

• Currently, India does not have an overarching guidance framework for the use of AI systems. Establishing such a framework would be crucial for providing guidance to various stakeholders in the responsible management of Artificial Intelligence in India.
• India currently does not have overarching legislation specific to AI. The closest to this is the draft Personal Data Protection Bill (2019) designed as comprehensive legislation outlining various facets of privacy protections that AI solutions need to comply with. The "Bill" was referred for examination and recommendations to a Joint Committee of both Houses of Parliament (called JPC) on 12 December 2019.
• Some key recommendations of the committee include regulating non-personal data in the bill, changing the threshold for reporting data breaches, regulating social media companies and intermediaries, the role of data protection officers (DPOs), and the fiduciary duties of hardware manufacturers.
• The most notable and contentious suggestions are about data localisation. Social media companies have expressed their concern that India’s data privacy bill requiring local storage and processing of data could increase the cost and complexity of delivering our services.
• National Mission on Interdisciplinary Cyber-Physical Systems (NM-ICPS): NM-ICPS is a Pan India Mission and covers the entire gamut of India that including Central Ministries, State Governments, Industry and Academia.
• It was launched in 2018 to be implemented by the Department of Science &Technology at a total outlay of Rs. 3660 crores for a period of five years. (NM-ICPS) will augment advanced research in next-generation technologies such as AI, Internet of Things, Machine Learning, Deep Learning and Quantum Information Sciences.
• Once the research and development ecosystem for AI takes a concrete shape with the implementation of the AI Strategy, technology research and development for AI and machine learning in cybersecurity will benefit immensely.

Concerns:

• Threat from the flawed analysis: While AI-assisted analytics could sometimes produce a real-time picture of a threat to national security, the choice of action to deal with it will require an application of human discretion taking into consideration aspects that a 'machine' was not equipped to handle.
• Threats and conflicts are creation of human beings and in both Intelligence analysis and decisions about 'action', human intervention will be a necessary and desirable paradigm. A flawed analysis or an unwise decision can cause grave harm to the nation.
• Keeping Humans in the loop: In the context of national security, the response part is totally governed by human decision-makers. It is possible that Command and Control might be in a position to delegate certain decisions in special circumstances to AI platforms out of compulsion of competition in time but it is unlikely that autonomous lethal systems, for instance, would be brought into play without putting 'humans in the loop'.
• The geopolitical environment around India: China’s AI program is considered ahead of India. China has made significant growth in dual-use technology. It is also enhancing its technical cooperation for AI with Russia. The trio (China-Russia-Pakistan) of the Sino-Russian relationship, the Sino-Pakistan relationship and the Russia-Pakistan relationship have been increased in the current geopolitical environment. India will have to assess these relationships to formulate an AI strategy.
• Security risks: Security risks in AI systems arise from their reliance on data and from its design and deployment environment. Some of these attacks are unique to machine learning systems and affect different parts of the machine learning development cycle. Adversarial machine learning attacks are designed to take advantage of vulnerabilities in the machine learning model with potentially harmful real-world consequences. Adversarial attacks affect the output of an AI system by introducing ‘perturbations’ to the input data.
  
  

  AI, like any technology, would tend to be transparent and its applications developed for the country's specific requirements will, therefore, have to be kept under wraps which is not an easy thing to do. Intelligence sharing and security cooperation among friendly countries are now tested on the willingness of either side to share the relevant technologies too. Fortunately, India is well placed to command the trust of its friends outside because of the deep convergence that had already been reached on national security issues with democracies like the US, Japan and Israel.

• Privacy Risks: AI systems rely on large amounts of training data and when an individual’s personal data is used there are bound to be considerable privacy concerns. Lack of adequate privacy safeguards may permit technology to wholly record and analyse an individual’s personal life without their consent or knowledge, significantly harming an individual interest by disregarding their preferences on the use of data. For example, using facial recognition technology for surveillance poses such risks. Such harm may be economic – stealing an individual’s credit card information, or emotional – where an individual’s personal details are the subject of public discussion.
• Increased instances of Misinformation: While analysis of Intelligence received from multiple sources is made easier and qualitatively superior, it has to be kept in mind that the capacity of the adversary to dish out 'misinformation' with the help of AI would also have increased.
• Intelligence sharing and security cooperation: AI, like any technology, would tend to be transparent and its applications developed for the country's specific requirements will, therefore, have to be kept under wraps which is not an easy thing to do. Intelligence sharing and security cooperation among friendly countries are now tested on the willingness of either side to share the relevant technologies too.
• Fortunately, India is well placed to command the trust of its friends outside because of the deep convergence that had already been reached on national security issues with democracies like the US, Japan and Israel.
• Economic constraints: Being a developing nation, India has economic constraints in its policies. Due to this, it cannot allocate resources as per the requirement of forces. Their security relies upon manpower. However, AI can do the work of multiple men at a quick pace. Thus, there is a need to invest in the research and development of AI platforms. AI platforms would be comparatively cheaper than manpower and expenses. It provides an opportunity to formulate a defence budget at a minimum expense.

Area of Improvement:

• Developing a policy to explore and aspects of securing borders with the integration of advanced technologies like AI.
• India can use heterogeneous robotic systems for surveillance of land and river borders by ALVs, UAVs and integrated systems, sharing data in real-time working as eyes and years of one single machine on the LAC.
• Procuring autonomous and remotely operated ground and aerial vehicles to increase the frequency of patrols. It will show the heightened presence of Indian patrols in the region, Which is important for showing dominance over the region.
• Working towards developing in-house technologies for virtual assistance in securing the borders like logistics, predictive analysis, and image and video processing.
• Focusing on establishing and developing institutions that will work on building robust algorithms of AI, according to current and future needs of securing the borders and national security.

Conclusion:

In both Intelligence evaluation and response, Einstein's famous dictum 'Imagination is more important than knowledge' has a great application – this is the human add-on to whatever Artificial Intelligence could provide to security. The imagination here is the ability to see beyond the available facts and hit the right interpretation and make the right option of action.

AI will be one of the most disruptive technologies of our generation and will change how we do business today, especially in the security sector. Even though AI is very helpful in a lot of aspects, it needs to be operated under human discretion when the tasks are dealing with human lives and the sovereignty of the country to avoid catastrophic level events. Given the complex terrain of navigating challenges posed by AI systems, it is integral that future deliberation, policy-making and regulation of AI is informed by multiple disciplines on an equal footing. These must be ethically, legally, technically and philosophically-informed throughout the process.