Cyber News

According to CNA, personal data of about 70,000 individuals was compromised after unauthorised access to a dataset created for vendor development and testing. The exposed information reportedly included names, NRIC numbers and past property addresses. This is not just another cybersecurity incident. It exposes a deeper problem in the way many organisations still think about data protection. The Real Problem Based on the public reports, the affected dataset was created for te

Lately at Semnet, we’re noticing clients asking different questions lately and adopting alternative brands and solutions. They’re not opening meetings with “Who’s the leader?” anymore. Instead, they’re saying things like: “What actually solves our problem with lower price?” For years, the safest move in enterprise IT was simple: pick the biggest vendor, sign the cheque, and hope for the best. Brand name equalled peace of mind. Less so nowadays. With tighter IT budgets while

Background The rapid adoption of AI across enterprises has introduced a new class of data risk. Unlike traditional systems, AI requires broad access to data in order to analyse, organise, and generate insights efficiently. In practice, this often leads to organisations exposing large datasets to AI systems without granular control. Traditional security controls such as encryption at rest, encryption in transit, and multi-factor authentication are necessary but insufficient. O

Practical controls to govern agentic AI in 2026, stop data leaks, contain CVE-driven exploits, and prove compliance without slowing innovation.