AI in cybercrime: what 28 records show

Strongest gains

1. AI chatbots improved threat detection and cut response delays, a review of 235 studies found

   A review of 235 peer-reviewed studies reported that pairing AI language models with scalable data pipelines improved detection accuracy and reduced response delays in threat detection and incident response.

   Systematic review or meta-analysis · Evidence: High · 100/100 · TRV-2026-0456

2. Greener AI network guard cut energy 43% and kept 98.7% accuracy in tests on public data

   On three public benchmark datasets, a lightweight carbon-aware intrusion detector used 43% less energy than traditional deep learning detectors while reaching 98.7% accuracy, and cut emissions 58.7% when the grid was high-carbon.

   Benchmark or lab evaluation · Evidence: Moderate · 52/100 · TRV-2026-1969

3. Trained only on normal traffic, an AI model caught network attacks with 93.20% accuracy in benchmark tests

   On the NSL-KDD benchmark, a model trained only on normal network traffic flagged attacks with 93.20% accuracy and a 2.13% false alarm rate, without labeled attack data.

   Benchmark or lab evaluation · Evidence: Moderate · 52/100 · TRV-2026-1656