Next-Generation Cybersecurity

AI & Deepfake Security

What it covers:

Protection of AI models against manipulation, data poisoning, jailbreaks and adversarial attacks; security of AI agents; detection of deepfakes and synthetic media; media forensics; and content provenance and watermarking for AI-generated material.

As artificial intelligence becomes embedded in business processes, AI systems themselves are increasingly becoming targets. Attackers can poison training data, manipulate model inputs or exploit weaknesses in language models and autonomous agents to trigger unintended actions. At the same time, generative AI makes it easier to create realistic fake images, videos and voices, which can be used for fraud, impersonation and disinformation. This field covers technologies that protect AI systems from attacks and help identify content that has been artificially generated or manipulated.

Why it matters:

Trust in AI systems and digital content is becoming increasingly important to the digital economy. AI & Deepfake Security is an emerging area of cybersecurity, with patent activity reflecting growing efforts to address risks that have gained prominence only in recent years.

Identity & Anti-Fraud

What it covers:

Passwordless and continuous authentication, decentralised digital identity and identity wallets, biometric liveness detection and anti-spoofing for face and voice, identity management for machines and workloads, and AI-based detection of payment fraud, account takeover and money laundering.

Stolen credentials and compromised identities are behind many successful cyberattacks. Next-generation identity technologies reduce reliance on passwords through cryptographic credentials and biometrics, verify users continuously based on behaviour, and give individuals greater control over their digital identities. As criminals increasingly use deepfakes and synthetic identities to bypass verification, liveness detection and AI-driven fraud analytics are becoming more important.

Why it matters:

As more transactions, services and interactions move online, reliably verifying who or what is on the other side becomes critical. This applies not only to people, but also to devices, software and AI agents, making identity security central to both cybersecurity and digital commerce.

Quantum & Hardware Security

What it covers:

Quantum key distribution, quantum random number generation, post-quantum cryptography and crypto-agility, as well as hardware-based security technologies such as physically unclonable functions, hardware roots of trust, secure boot and firmware attestation, and protection against side-channel and physical attacks.

Cryptographically relevant quantum computers could break many of the public-key cryptographic methods used to protect digital communications today. This creates a risk that data intercepted now could be decrypted in the future. Post-quantum cryptography develops algorithms designed to withstand such attacks, with NIST publishing its first post-quantum cryptography standards in 2024. Quantum key distribution offers a different approach, using quantum-mechanical properties to help secure key exchange.

Security also increasingly relies on protections built directly into hardware. These include chips with unique physical characteristics, trusted boot processes and countermeasures against attacks that extract secrets through power-consumption measurements or deliberate hardware faults.

Why it matters:

The transition to quantum-resistant cryptography will require extensive changes to digital infrastructure, from communication networks to connected devices. Hardware security provides an additional layer of protection, helping establish trust in the systems on which software and digital services depend.

Privacy & Confidential Computing

What it covers:

Homomorphic encryption, secure multi-party computation, zero-knowledge proofs, differential privacy and privacy-preserving machine learning, as well as confidential computing based on hardware-protected trusted execution environments.

Conventional encryption protects data while it is stored or transmitted, but data generally needs to be decrypted for processing, potentially exposing it to unauthorised access. Privacy-enhancing technologies address this gap in different ways. They enable computations on encrypted data, joint analysis across organisations without directly sharing sensitive information, or verification of specific facts without revealing the underlying data. Confidential computing uses hardware-protected environments to isolate data and AI models from other software running on the same system, including, in some configurations, access by cloud operators.

Why it matters:

Organisations want to use sensitive data for AI, research and business collaboration, but privacy requirements, regulation and commercial concerns often limit what they can share. Privacy-enhancing technologies make it possible to extract value from data while reducing exposure and retaining greater control over its use, creating opportunities in healthcare, finance and cross-border data collaboration.