Artificial Intelligence reshapes cybercrime, enterprise operations and security services

Researchers at Sophos warn that cybercriminals are increasingly adopting generative AI for social‑engineering attacks, custom phishing, voice‑phishing bots and even AI‑assisted malware, while some operators fear the technology could render traditional illicit services obsolete.

A F5 survey shows that 78 % of companies now run AI inference in production, typically with seven models concurrently. Executives cite gaps in governance, security and infrastructure as the main challenges, noting the need to treat AI workloads like any critical application, to secure prompt and token layers, and to manage autonomous AI agents with dedicated identities.

BlueVoyant has launched a Microsoft‑focused security service that inventories, governs and protects AI agents such as Copilot, providing identity controls, threat detection and data‑loss prevention within Microsoft Entra, Defender and Purview.

The Austrian Society for Artificial Intelligence is hosting the European Summer School on AI in Vienna, gathering over 400 international researchers to promote "digital humanism" and trustworthy, explainable AI.

Kaspersky’s latest dark‑web analysis reveals that credentials from small and midsize enterprises now account for roughly 60 % of listings, with attackers using these accesses for ransomware and data theft, underscoring the growing targeting of smaller firms.

Across these developments, AI emerges as both a catalyst for new threats and a driver of operational transformation, prompting organisations to strengthen governance, security architectures and workforce preparedness.