Three vulnerabilities in Salesforce Agentforce, known as SalesBleed, enabled attackers to hijack AI agents for data theft and phishing attacks.
A new Android malware named RemControl is targeting banking customers across Europe, the Middle East, and Canada by abusing Accessibility Services and AI-generated code.
The 'SalesBleed' attack chain targets Salesforce Agentforce, allowing attackers to use prompt injection to silently exfiltrate sensitive CRM data via zero-click methods.
AI allows attackers to rapidly iterate through failed attempts, requiring Security Operations Centres to move toward a stateful model of shared operational memory.
A new report from Anthropic details how AI agents are being used to industrialise cyberattacks, propaganda, and surveillance workflows.
