PromptSteal, developed with help from the Hugging Face AI community, has been used by Russian intelligence to infiltrate Ukrainian networks, enabling remote malware deployment and control.

AI-powered malware is rising, enabling on‑demand generation and rapid deployment that could redefine cyberwarfare and strain traditional defenses.

Mobile and IoT security remain precarious, with a 2025 threat report showing a 67% jump in Android malware, millions of infected devices, and many strains slipping past Google Play filters.

PromptFlux uses Google's Gemini AI to adjust behavior in real time, boosting its ability to evade security controls in changing environments.

Experts urge robust defenses, including AI-enabled detection, behavioral and anomaly analytics, automated response, and continuous security training.

IoT threats persist, with botnet-driven DDoS risks driven by Mirai, Gafgyt, and related vulnerabilities.

The overarching takeaway is clear: AI-augmented cyber threats are evolving quickly, demanding advanced, proactive defense across individuals, organizations, and critical infrastructure.

Two notable AI-driven strains, PromptFlux and PromptSteal, leverage large language models to adapt and evade detection, complicating containment after deployment.

Threat intelligence organizations report AI-powered malware being used to target Ukraine, with allegations that Russian forces deploy AI-generated malware against critical infrastructure.