AU10TIX has issued a stark warning about the accelerating threat of agentic AI and emerging quantum risk, declaring 2025 the “Year of Machine Deception” in a new global identity fraud report released today.
The special edition of Signals for 2026 details how identity fraud has evolved from isolated, repetitive attempts into adaptive, self-learning ecosystems powered by Fraud-as-a-Service and automated feedback loops. AU10TIX’s research shows that what once appeared as unrelated “repeaters” across platforms were actually early rehearsals for coordinated attacks designed to refine themselves over time.
As agentic AI systems become capable of independently iterating, improving and redeploying synthetic identities, the report argues that fraud is no longer a set of discrete events but a living, self-optimising system. AU10TIX says this marks a turning point for global digital risk, as machine-generated deception becomes faster, more scalable and increasingly difficult to detect with traditional methods.
To counter this, AU10TIX has developed an early-warning framework designed to identify the moment “truth begins to drift.” By analysing behavioural, biometric and metadata signals across billions of identity events, the system detects when small anomalies begin repeating across networks and forming a coherent pattern. The company reports a 97.5 per cent correlation between these early behavioural irregularities and confirmed fraud, demonstrating that what once looked like random noise is now a measurable precursor to coordinated attack activity.
“Fraud is no longer a static event; it’s a living signal moving through networks and devices,” said Yair Tal, CEO of AU10TIX. “Our mission is to protect customers not just by responding to attacks, but by anticipating them. Our early-warning system helps ensure their businesses stay one step ahead, detecting risk before truth starts to drift.”
The shift toward predictive assurance is already delivering tangible results. After AU10TIX introduced real-time anomaly scoring in April 2025, customers saw a 72 per cent reduction in selfie-injection deepfake attacks by August, marking a significant advance in combating machine-driven deception.
Looking ahead, the report identifies two major fronts shaping the 2026 threat landscape: the rise of agentic AI fraud engines capable of autonomously creating and adapting synthetic identities, and quantum risk, which threatens to undermine the mathematical foundations of current encryption standards.
In response, AU10TIX has launched a Predictive Resilience Framework that integrates anomaly intelligence with quantum-resilient cryptography across verification events. The model combines three interconnected layers — hash, encrypt and predict — to protect both the mechanics and mathematics of digital trust. Hashing secures signals against tampering, post-quantum-aligned encryption safeguards data from emerging decryption threats, and predictive analytics surface the earliest indicators of AI-driven or behavioural spoofing before attacks scale.
The report also highlights several leading indicators for 2026, including a projected 100 per cent increase in presentation spoofing, a 60.7 per cent rise in identity drift and a 36.4 per cent increase in credential replay attacks.
As machine-driven deception accelerates, AU10TIX says its predictive resilience architecture is positioned to help organisations stay ahead of threats that learn, adapt and evolve as quickly as the technologies they exploit.
