OpenAI Addresses Ongoing Prompt Injection Attacks on Atlas AI Browser

OpenAI Addresses Ongoing Prompt Injection Attacks on Atlas AI Browser

First seen 23 Dec 2025, 19:00 UTC •ScworldDigitaltrendsCyberscoop •31.7

Article Content

Browse articles
ThreatCluster

OpenAI has reported that prompt injection attacks targeting its Atlas AI browser continue to pose a significant security challenge. These attacks manipulate AI agents through concealed instructions in web pages or emails, prompting the company to enhance its defenses against these threats. OpenAI acknowledges that completely eliminating such vulnerabilities is unlikely.

How this analysis works

Extracted Entities

1 / 2