Apple patches dyld flaw linked to targeted attacks
Apple said the memory corruption issue may have been used against specific individuals.
Capvise Media / Stories
Security
Privacy, identity and trust in a connected world.
110 storiesBrowse by month
Microsoft’s Cyber Pulse report puts agent governance under scrutiny
The February report argues for treating agents as accountable identities.
Microsoft finds promotional instructions inside AI-summary links
The February research describes attempts to bias future recommendations.
Microsoft research tests how fine-tuning can undermine model safeguards
The February study concerns post-deployment training, not a universal chat prompt.
A safe prompt cannot repair an unsafe agent dependency
Microsoft’s January reports describe risks below the conversation layer.
Microsoft survey finds AI controls lagging adoption
The January report draws on more than 1,700 security leaders.
Microsoft describes an AI workflow for finding detection gaps
The January account keeps analysts responsible for validating the result.
WhatsApp adds stricter settings for people facing targeted attacks
The optional mode restricts attachments and calls from unknown contacts.
Android strengthens theft lockouts and recovery checks
January changes separate phone access, identity checks and remote recovery.
Microsoft issues emergency Office fix amid exploitation
The January response included a Word security fix with specific installation requirements.
SharePoint phishing campaign shows why password resets can fall short
Microsoft describes compromised sessions and malicious inbox rules in an energy-sector campaign.
An AI agent inventory needs permissions, not just model names
Microsoft’s January guidance focuses on the connections that turn an assistant into an actor.