Cybersecurity continues to be a critical priority for organizations and nations, driven by increasingly complex landscape shaped not only by expanding attack surfaces and sophisticated threat actors but also by the rapid advent of artificial intelligence (AI) technologies and the advent of quantum computers. The rise of large language models (LLMs), generative AI, agentic AI and chatbots not only enhance defensive capabilities but also expand the attack surface and introduce novel risks.
The widespread adoption of AI tools creates new vectors for data leakage, social engineering, and automated attacks, demanding vigilant governance and controlled usage within organizations.
AI-driven tools such as LLM-powered chatbots, automated code generators, agentic AI introduce unique risks, from inadvertent exposure of sensitive data during AI interactions to malicious actors using AI to automate sophisticated phishing, ransomware, and zero-day attacks. Organizations must incorporate controls to regulate and monitor how AI tools are deployed internally, ensuring these do not inadvertently increase vulnerabilities.
AI is a double-edged sword, while AI significantly improves threat detection, response, and efficiency in cybersecurity. AI systems are themselves vulnerable to unique attacks. Effective AI security requires securing data, AI models, and usage throughout the AI lifecycle, protecting against threats like data poisoning, adversarial attacks, and model theft.
Quantum computers pose a serious threat to current encryption standards like RSA, ECC, and SHA-256, which could be broken within minutes by powerful quantum algorithms. To counter this imminent threat, cybersecurity strategies must prioritize Post quantum cryptography (PQC) and should include continuously assessing cryptographic infrastructure for vulnerabilities.
As we seek innovation within the Unisys Innovation Program, embracing security not only aligns with the modern threat landscape but also showcases our commitment to safeguarding sensitive information and ensuring a resilient cybersecurity posture.
Students are encouraged to brainstorm innovative ideas that incorporate the listed areas and technologies, or to explore unconventional approaches and develop your own creative solutions.
| Identity and Access Management (IAM) | Vanilla Cybersecurity | AI for Cybersecurity | Cybersecurity for AI |
| Multi-Factor Authentication (MFA) | Intrusion Detection and Prevention Systems (IDPS) | Self-learning AI detecting network anomalies | Adversarial training frameworks |
| Single Sign-On (SSO) | Security Information and Event Management (SIEM) | Phishing and deepfake detection | AI model theft detection and mitigation |
| Identity Providers (IdP) | Endpoint Detection and Response (EDR) and Extended Detection and Response (XDR) | AI supplementing endpoint detection and response (EDR) | Retraining to counter model theft and bias |
| Privileged Access Management (PAM) | Red Teaming for improving resilience | Predictive threat intelligence | Model Integrity & Poisoning Defense |
| Incident Response & Forensics | Enhance Zero Trust by enabling real-time behavior analysis |
Here are a few scenarios that you may wish to consider.
- AI Governance Controls for Safe and Compliant Use of Large Language Models and Chatbots in Enterprises
- AI model theft detection by watermarking AI models
- AI-Powered Phishing and Deepfake Detection to combat evolving social engineering attacks
- Secure Development and Deployment Pipelines for AI systems, including data poisoning mitigation
- Cloud and Hybrid Environment Unified Policy Enforcement including AI-driven compliance monitoring
- Endpoint Detection and Response (EDR) systems enhanced with AI for rapid malware identification
- Secure Supply Chain Management leveraging blockchain and AI risk scoring models
- User Behavior Analytics to identify deviations indicating potential credential theft or lateral movement
- Cybersecurity measures must ensure data and model integrity since data poisoning and manipulation of model outputs are used by attackers
- Fraud detection with AI - AI excels at spotting patterns in financial transactions or login attempts that deviate from normal behavior