Generative AI and human entropy
2024 marks year zero of a frightening new era for cybersecurity: the infusion of Generative AI into botnets. Old-generation scripts that clicked mechanically have been replaced by neural networks trained to imitate human chaos.
Ad-Fraud Damages (2024)
Source: Market estimates
Beyond Scripts: Simulating Entropy
Until 2023, anti-fraud filters analyzed behavioral perfection: if the mouse moved too geometrically, traffic was flagged as invalid (IVT).
Today, thanks to LLMs, bots implement entropy: irregular micro-movements, hesitant scrolling, and even simulated typos in forms.
Why Software is No Longer Enough
- Outdated Behavioral Models: AI bots generate calculated chaos mathematically indistinguishable from a distracted human.
- Adversarial Machine Learning: Bots study anti-fraud filters and adapt in real-time.
- Dynamic Profiles: Creating fake users with credible browsing histories to attract expensive ads.
The Death of Behavioral Filters
Faced with the algorithmic impossibility of distinguishing an AI-bot from a human using only data, the software defense castle collapsed, exceeding 88 billion in damages.
The only logical conclusion was reached: without an external 'physical' block (Offline), online fraud can no longer be solved.