Cybersecurity

Microsoft concluded 2025 with 1,129 total patches, an 11.9% increase over 2024, as attackers continue to exploit zero-day vulnerabilities in Windows and legacy protocols. AI security remains fragile, with new research showing poetic prompts bypass safety filters with 90% success, while privacy concerns in healthcare are driving patients to disconnect vital monitoring devices. Notable Insights: • Poetic prompts can bypass the safety mechanisms of 25 frontier AI models with success rates exceeding 90%, proving that simple stylistic changes can defeat current alignment methods. • 34% of seniors reject vital healthcare IoT devices due to privacy fears; despite being technically capable, they lack confidence in the apps' opaque data-handling policies and lack of breach notification protocols. • Microsoft patched 1,129 vulnerabilities in 2025, marking an 11.9% increase from 2024 and the second consecutive year the company has exceeded one thousand patches.