AI Browsers Raise Security And Privacy Concerns
McAfee warns that agentic AI-powered browsers like ChatGPT’s Atlas and Perplexity’s Comet actively automate tasks but pose evolving security and privacy risks. McAfee's research notes exploits ranging from data theft, Gmail/Drive access, malware, to injected malicious memory, though some attacks have become harder as vendors add guardrails. It recommends restricting permissions, disabling model-learning, using logged-out modes, and monitoring privacy policies to reduce risk.
Key Points
- 1Describe agentic AI browsers summarizing pages, automating tasks, and proactively acting like personal assistants.
- 2Warn about exploits including data theft, Gmail/Drive access, malware, and injected malicious memory across sessions.
- 3Advise practitioners to restrict permissions, disable model learning, use logged-out mode, and monitor privacy policies.
Scoring Rationale
Practical, vendor-backed guidance with clear mitigation steps; limited novelty beyond existing security best practices for users.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems