Proofpoint Targets Email Attacks on AI Assistants
Cybersecurity company Proofpoint has announced new protections against email-based attacks targeting AI assistants including Microsoft Copilot and Google Gemini, as organisations increasingly integrate artificial intelligence agents into workplace workflows.
The firm revealed four security innovations designed for what it terms "the agentic workspace" – environments where employees collaborate alongside AI agents and assistants.
Attackers are embedding malicious prompts in emails to manipulate AI assistants, using prompt injections to provide false information to users, confuse AI-based defences, and steal sensitive data, according to Proofpoint's announcement.
The company's Prime Threat Protection solution will block these exploits before they reach inboxes, enabling staff and AI agents to trust workplace interactions.
Proofpoint's Data Security Complete solution addresses growing data risks in AI-enabled workplaces by providing discovery, classification and control across endpoints, email, web and cloud platforms.
The solution includes Autonomous Custom Classifiers for dynamic data classification and creates consolidated risk maps showing cross-channel data lineage and potential exfiltration risks.
A companion AI Data Governance capability enables organisations to discover both authorised and unauthorised AI usage while applying policies to prevent data exfiltration and privacy violations.
Securing AI Agent Activities
The company's Secure Agent Gateway, built using Model Context Protocol, monitors and controls how customer-deployed AI agents access organisational data.
The gateway enforces data usage policies and can block or redact sensitive information before sharing between agents or with human colleagues.
Proofpoint Satori Agents operate within the company's security platform to handle data loss prevention alerts, recommend phishing simulations, and resolve user-reported email threats automatically.
"The agentic workspace is here and one of the most profound shifts in terms of how work gets done," said Proofpoint CEO Sumit Dhawan.
“Protecting the agentic workspace is the next evolution of human-centric security, extending beyond people to safeguard AI agents and the points where they collaborate and share data.”
The AI exploit detection capability for email is expected in Q4 2025, while Data Security Complete is available from Q3 2025. The Secure Agent Gateway and Satori Agents enter phased availability from 2026.