The rise of Artificial Intelligence and autonomous systems is dubbed as the Revolution in Military Affairs, whereas analogously, in the realm of intelligence affairs or the intelligence community, this is perceived as Revolution in Intelligence Affairs. Innovations and advancements in computing and information systems are transforming the conventional designs of the war, intelligence and national security. With the rapidly changing research and technology, the Intelligence Community has spearheaded new initiatives to incorporate evolving technologies such as cloud, artificial intelligence (AI), and deep learning etc. that will eventually revolutionize intelligence affairs.
Here we will focus on the modernization campaigns and prospects of expediting and remodeling the intelligence affairs by using AI (machine learning, neural networking) and Big-Data.
Starting from data-driven intelligence in modern times, the first change in intelligence affairs is the huge influx of data. ‘Data’ has been ranked highest commodity taking over ‘oil’ and it is in terms of value attached with data analytics, especially big-data analytics. On one hand, there is no denying that today, we are not in a phase where we can exploit the torrent of data to their optimum level for national security, because we are not able to comprehend or makes sense of it. On the other hand, if a researcher has to conduct nationwide quantitative research in this modern internet era where data is digitized and streamlined to an extent that it could be possible to run algorithms on a quantum computer to get very complex results out of that gigantic data-set. However, with the help of machine learning and algorithms, there are possibilities to make sense of big data more efficiently, use it for intelligence, and counterintelligence. Analysts are calling it something that can help us now to find a needle in the haystack but when it comes to readiness its employment by the defence and intelligence agencies can lead to spectacles.
To find the real-time examples where Big-data analytics has been used by the intelligence agencies, we have to look at the technologically advanced nations where the US is a leading one. The usage of emergent technologies like Big-Data analytics through machine learning is now being adopted by the National Security Agency of the United States for surveillance and reconnaissance. Big data sets are analyzed to draw patterns and coefficients for security/intelligence assistance using AI algorithms. Network-based surveillance programs like the US National Security Agency’s ‘Prism’, or something similar to the Israeli Defense Forces ‘Pegasus’ or China’s manifestation of AI in police goggles for face recognition are the practical examples where nations use AI for defence and strategic purposes.
Similarly, in the case of the UK, Government Communication Headquarters (GCHQ) researchers go to great lengths to clarify why AI holds the secret to improved national security in a recent paper titled “Pioneering a New National Security.” The amount of data that GCHQ works with puts intelligence services and law enforcement authorities under a lot of stress, according to the organisation; AI could help alleviate the strain by improving not just the pace but also the consistency of expert decision-making. GCHQ is now actively engaged in programmes including artificial intelligence. While the organisation would not reveal the specifics of its use of the technology, Fleming mentioned several collaborations with AI-related startups around the world, as well as a strategic alliance with the Alan Turing Institute, which was established to advance AI and data science research.
“AI, like so many other technologies, has enormous potential for societal progress, stability, and defence. Many of our missions already depend on AI to protect the nation, its inhabitants, and its way of life. It enables our talented analysts to handle massive amounts of complicated data and enhance decision-making in the face of extremely complex challenges, ranging from child protection to cybersecurity,” said GCHQ director Jeremy Fleming.
AI-assisted Intelligence Analysis
In today’s information-driven world AI has become a key element for decision-making and playing an extremely important role in the administration as well. Although the technology is in its nascent phase still the AI-assisted intelligence processing is proving to be valuable because it helps saving time and resources in deriving information from unstructured and incongruent data sets, rendering more reliable results with fewer chances of anomaly and avoid collateral intrusions.
A report published by the Rand Corporation categorizes the AI-assisted intelligence analysis into three segments:
- Cognitive Automation: the machine replication of human sensory processing would significantly reduce the time needed for human operators to interpret large volumes of data, while also potentially reducing intrusion by minimising the volume of content that is subject to human review. According to the current developments, AI is useful in object classification and identification, such as spotting a person in a crowded area or a vehicle. Similarly, the current research in the field of AI and intelligence affairs focuses on video summarization where through machine learning a summary of the video will be generated which may include the context as well. According to the Software Engineering Institute (SEI), “the long-term goal would be to recognize and search for patterns of life across multiple videos, with the ultimate goal of predicting future activities and events”.
- Filtering, Flagging, and Triage: As we know that the working mechanism of intelligence agencies is that they collect the data/information from different sources, then filter out the wanted part, and prioritize the rest. However, if AI is used in this system, it will bring an imperative change in the process because it will save time, human resource, collateral intrusion, and most importantly will spot anomalies and patterns much better than humans. For extracting intelligence from bulk data, AI is likely to be most useful when deployed as part of an interactive ‘human-machine team’ analysis workflow.
- Behavioural Analytics: It is the application of complex algorithms to individual-level data for analyzing personalities and predicting human behaviours. The application of AI-assisted analysis for behavioural analytics by the intelligence community will help in detecting an insider threat or a threat to the public; also, it can help in identifying the potential intelligence sources and recruitments conducted by the agencies. There is an ongoing debate, related to the ethics and privacy of individuals if this technology has been implemented on a broader scale but no one can deny the real-time benefits of behavioural analytics when it is used for stopping crime and countering terrorism. However, recent research into the prediction of life outcomes using a mass collaboration approach concluded that ‘despite using a rich dataset and applying machine-learning methods optimized for prediction, the best predictions were not very accurate and were only slightly better than those from a simple benchmark model’.
Computers and network of networks (internet) have set the foundation of a so-called ‘information revolution’ and transformed the social and economic modus operandi. Because the internet was invented for military use, thereby its usage and implications in military and intelligence affairs are still the bone of contention. Especially in modern times, when the militaries are functioning on the doctrine of ‘jointness’ where the Army, Navy, Air Force work together in a battle theatre using Joint Situational Awareness Systems. Similarly, now a day, the warzones have become a virtual game for the decision-makers because it is being monitored through advanced command, control, computers, and intelligence, surveillance and reconnaissance systems also known as C4ISR. This system works on the network of networks, which is commonly known as the internet or cyberspace.
Cybersecurity has become a new frontier of national security and with the rapid digitization of data and connectivity of critical infrastructures through cyberspace; an ever-growing threat now underpins every part of society. Here it can be a kinetic or a non-kinetic threat to the national security, such as it can disrupt the physical services of power production, water treatment, sabotage banking systems etc. whereas at the same time a non-violent cyberattack can result in creating an environment of hostility through spreading disinformation on social media. In both cases, nation-states and instruments of national security, like intelligence agencies, can take advantage of AI to secure their national cyberspace and can pre-empt any such malicious activity in their cyberspace before it happens.
We can again quote the example of the US; the country holds the greatest cyberspace security offensive/ defensive capability –more than any other nation in the world, in both government and commercial aspects –yet the US over the past one decade suffered from systematic vulnerabilities- the cyber-attacks and cyber intrusions of enemies. This is why the new strategy of the US is called “defending forward and persistent engagement,” where defending forward is a defensive strategy with offensive posture adopted to disrupt malicious actions and its sources preemptively. It also falls under the ambit of the intelligence community because it is not deterrence by denial or deterrence by counterforce but it is an idea of getting into enemy’s cyberspace, like sneaking in or eavesdropping, and neutralizing them before they can attack. This strategy is intended to be proactive, observed, pursue, and counter the cyber operations of US adversaries before they occur.
In the case of the UK, nearly half of UK companies have experienced a cyberattack in the last year, with a quarter of those resulting in a substantial loss of money or data. AI could help the department properly detect malicious malware and keep its dictionary of documented trends up to date to predict potential attacks. By automatically fact-checking posts and weeding out botnets and troll farms on social media, the tool could be used to combat web misinformation and deep fakes.
When we talk about the intelligence agencies, the first thing that comes to mind is the 1960’s concept of espionage and counter-espionage, which has changed much from the concept of James Bond to computer hacking and stuff. There are two types of insider breaches; one is intentional while the other is unintentional. The concept of an insider threat has also changed because previously, one has to steal information through taking pictures or scribbling notes etc. but now one can carry them on a USB stick or mail them in an encrypted form etc. Such threats could be mitigated through AI-led cybersecurity systems or regular analysis of the security performance, especially cybersecurity, through data-driven rating systems. There are numerous such systems readily available, which can collect data regularly from hundreds of sources and then mapped according to the external infrastructure of the organization, which needs to be evaluated from the algorithms and technical experts. It plays an important role in the awareness component, allowing us to position the risk factors, and it plays an important role in agility. The next step is usually the filtering process, where based on the various risk vectors indicating the compromised systems, diligence information and user behaviour breaches etc., the system calculates and rates the cybersecurity of the critical infrastructure.
AI with machine learning goes beyond big data to create a complete incidence response system that is much more effective. Considering its application in a cyberattack, AI and ML can help us create better attack and defence models that were simple, not possible before. A defensive AI in cyber can help us in creating new signatures every 5 or 10 minutes to avoid cyber invasion and the attack that the antivirus signatures were not able to respond to. It also assesses different aspects associated with behaviour response. A practical manifestation of this defensive AI is the Socket Security (SOCKS) operations centre.
On the attack side, machine learning can help emulate the same visual, text, and audio to impersonate someone, it can assist in the decryption of sensitive data much faster, and brute force attacks or denial of service attacks etc. In the realm of cyber, AI acts as a dynamic weapon, which can be used to create more advanced cyber-attack scenarios and vectors. We have seen the birth of the new technology that is infecting many of our social media that we sometimes refer to as Deep Fakes. One can affect the impact of elections or even other important events by exploiting AI technology as a coercive tool.
In the context of intelligence gathering from cyberspace, AI plays a blazing role because it can help in espionage where one can gather secretive information or fetch large databases for analysis and helping in situational awareness etc. Similarly, AI will include technologies like fuzzing which helps us to populate the narrative high-risk areas that privately or not previously identified as a potential risk against a specific object like a critical infrastructure object. Moreover, it can be pivotal in the training and awareness as well because AI can help us decide that what is the more effective training, which can produce the best results, and what are the areas that need to be addressed to have a complete solution in the environment.
The automation of social engineering attacks is another potential threat. With the help of the application of AI in data sciences, one can anticipate other’s actions and mould their perception. By collating a victim’s online information, attackers can automatically generate malicious websites, emails and links that are custom-made for clicks from that victim (sent, for example, from addresses imitating their real contacts). Our online activity will be used to identify patterns of our personality in future and employees will be graded based on their online performance. Further developments in this area could see chatbots gaining human trust during longer and more creative online dialogues. In the time-sensitive context of an election, Deep Fake and disinformation campaigns on social media and the internet is a weapon that amplifies the security dilemma and can result in catastrophic consequences.
Technologies like Blockchain, which adds extra elements of stress that simply were not possible before and this is an element of trust without a central authority. Other technologies playing a significant role as AI are technologies like Quantum Computing, which can be used for cybersecurity, in defence, it can help us create Quantum encryption, which creates a one-time scenario almost making encrypted our unbreakable cryptology. Similarly, on the cyberattack side, Quantum Computing will be employed to break keys much faster than will simply possible before.
In short, our world is going through digitization and soon AI will act as a fuel for cyber warfare, as we already know about the advancement AI brought to the data sciences and in its connection with cyberspace, now an actor can assess every click and depict patterns of our cyber interaction.
To conclude, it is pertinent to mention that the above-mentioned emergent technologies, especially Artificial Intelligence (AI) possess huge potential to enhance the intelligence work carried out by the national intelligence community. Intelligence agencies can harness the potential of these technologies but to take full advantage of these possibilities, there is a need for standardized procedures for designing, testing, and assessing new AI tools in their operational context. Many applications would be uncontroversial if they merely reduce the time and effort taken to process vast amounts of data that may have been processed using less reliable manual approaches previously. Although there are myriad ways of using the technology by the intelligence agencies such as using it for identification, data gathering, decision-making, situational awareness, analytics etc. at the same time there are concerns of citizen’s privacy, human rights and ethical considerations, which should be taken in the account before the development of such systems. Concerns regarding the ethical use of AI are highly subjective and context-specific and if such a system is deployed then there should be a periodic analysis and re-evaluation of the need and proportionality of any possible interference, the training data used to construct a model, and the decision-making mechanism. It is palpable that the use of AI raises a host of new concerns, implying that more regulation and clarification is required to ensure that AI research capabilities are used ethically and responsibly, taking into account topics like need and proportionality, openness and responsibility, and collateral intrusion risk.