Artificial intelligence (AI) and machine learning (ML) technologies have systematically changed the landscape for many industries. From bringing new innovative automated processes to the table to keeping companies at record-high efficiency, the rise of AI has been astronomical.
The ethical conundrum surrounding widespread, unsupervised AI/ML use still lingers, however. Not only does AI threaten human livelihoods and employment prospects, but it also amplifies the cyber threat landscape, something which has simultaneously grown and expanded as the world becomes increasingly more digitised.
Highly sophisticated and intuitive tools backed by AI/ML technology are being leveraged by malicious actors trying to execute attacks or extort money or information. At the same time, AI/ML allows cybercriminals to refine and expand their attack methods and make them exponentially more covert and dangerous.
However, while the cyber threat landscape has grown, AI/ML also presents organisations with critical new defence strategies and incident response solutions to protect systems and data from emerging threats.
The guidance below will provide a top-level overview of the uses of unethical and authentic AI and what the future looks like with the technology only poised to grow even more influential and available in the coming years.
AI/ML powers a new generation of cyber threats
Malicious actors are highly resourceful even without the augmentation of AI for their ulterior motives. While it’s difficult to find recent specific examples, security experts have pinpointed a handful of cases in which AI/ML can be weaponised by cybercriminals.
- Automated spear phishing and social engineering attacks: AI makes it easier to conduct highly convincing and deceitful phishing campaigns via email, SMS, and web-based platforms. Cybercriminals can exploit publicly available data to develop and launch calculated AI-backed attacks that exploit users or deceive them into divulging sensitive information innocuously.
- Enhanced and rapid malware or ransomware generation: Generative AI – i.e. the tech that underpins ChatGPT and similar tools – is capable of generating lines of code within moments. As a result, code can be adjusted to create malicious software, which can then be unleashed on users through phishing or brute force. Malware that ends up in a company’s network(s) can disrupt operations and lock out users until a ransom is paid. Even highly secure networks like the UK Government are at risk.
- AI-enabled botnets: Botnets have been empowered with the help of AI to enhance their evasion techniques while bolstering their attack methods. AI-powered botnets can analyse network behaviour and adapt their attack patterns accordingly. The bots’ intelligent reconnaissance capabilities can probe the systems most vulnerable and build a comprehensive picture of a business’s attack surface.
- Convincing deepfakes: AI’s influence in creating deepfakes or fabricated videos or photos has been widely criticised, and rightly so. However, fake videos and audio ‘recordings’ are likely to proliferate when helping bad actors prey on unsuspecting victims or those who lack vital cyber knowledge. Furthermore, deepfakes impersonating notable public figures or executives can trick employees into performing actions that they otherwise would not authorise.
Fighting Back with Defensive AI
While AI is undoubtedly expanding the threat landscape, it’s reassuring to know that cyber security enterprises and experts are combining efforts to find ethical and legitimate uses for AI as far as improving cyber prevention and defence solutions.
AI’s fundamental use and ‘raison d’etre’ is to automate repetitive, manual tasks and make humans more productive. AI’s applications and capacity for working autonomously far outweigh that of humans, hence why tasks like data collection and analysis, system management, vulnerability detection, and others can be entrusted to intelligent algorithms.
By empowering humans and reducing their alert fatigue, they can dedicate more resources and time to tasks that warrant deeper investigation, while AI can do most of the arduous groundwork to enable informed decision-making. Humans are freed up to focus on securing customer-client financial transactions and complex auto-enrolment payment schemes. Unlike AI, businesses can foster trust to run operations and safeguard integrated e-Commerce platforms like order processing systems across various industry settings.
Cyber security teams are adopting enterprise-level, ‘white hat’ AI solutions to bolster specific facets of their defence strategies. AI makes established cyber threat detection and prevention systems far more effective in the following ways:
- Enhanced threat detection: AI can analyse complex networks and interconnected systems to establish benchmarks for activities that can be considered ‘safe’ or ‘normal’. Anomalous incidents, such as unrecognised file uploads or network activity can then be identified and highlighted to teams. Bilaterally, AI can examine various factors like file characteristics, code patterns, and lateral movement to determine if files or scripts entering the incumbent system are safe or dangerous.
- Predictive security analytics: AI tools can aggregate innumerable security logs and raw incident data in mere moments; comparatively, human security analysts would take hours collating and visualising the important top-level data to present and share. AI alleviates the potential for human error when identifying trends and malicious activity patterns.
- Threat intelligence: AI can autonomously gather security-related information across a broad spectrum, from open-source repositories to the dark web. Cyber security solutions that have successfully embedded and integrated AI can use compiled data to identify emerging threats, assess risk exposure, and correlate indicators of compromise.
- Threat hunting and management: Using AI, cyber teams can automate the process of finding vulnerabilities in networks, endpoints, and web applications. With the help of ML algorithms, network traffic, logs, and data sources can be continuously and methodically monitored, with AI tools executing required actions based on predefined cyber security criteria and benchmarks. The most critical vulnerabilities or threats can be prioritised for timely and proactive containment.
An ethical imperative for AI security
As AI/ML proliferates on both sides of the cyber security landscape, the ethical application of these technologies has become paramount.
To reinforce the main talking point of this guidance, cyber defence teams must ensure any integrated AI systems are engineered responsibly and ethically, with all controls robustly assessed for misuse or unintentional harm.
Other focus areas include (but are not limited to):
- Guaranteeing fairness, accuracy and explainability in algorithm decisions
- Avoiding the proliferation of discrimination or misinformation
- Maximising transparency while preserving confidential data practices
- Enabling human oversight and control over all AI-enabled systems
- Objectively analysing and assessing AI tools’ propensity for unconscious biases
- Planning for potential adverse consequences and mitigation procedures
The future of AI in cyber security
Cyber security experts – both third-party vendors and in-house teams – can evidently use AI to improve multiple aspects of their security posture. However, establishing any AI cyber defence solution is never a one-and-done exercise. Threat actors constantly leverage AI and other technologies to launch attacks at incredible speed while diversifying their attack methods.
The more threat actors use AI to conduct adversarial attacks on established AI systems, the more likely that legitimate businesses risk falling behind in the growing battle to uphold cyber security and data integrity. Security experts must remain abreast of the developing threat landscape and emerging trends, and adapt their strategies and security controls to circumvent attacks and minimise their attack surface.
The only way to truly combat the growing problem of malicious AI is to leverage ethical and legitimate AI. Instead of viewing the technology as the underlying enemy, it’s crucial to leverage its capabilities in the correct ways, providing the right level of augmentation and support for security analysts. Organisations that can demonstrably and strategically adopt AI stand a much better chance of keeping sensitive data secure and preventing it from falling into the wrong hands. As a result, public trust will remain more resilient and unified than ever, particularly in a digital age where transparency and accountability remain more important than ever.
Try Phishing Tackle’s managed intelligence platform for customisable and automated security training simulations designed in response to the modern AI-enhanced threat landscape. Request a demo today to evaluate capabilities aligned with your organisational needs.