An ethical hacker demonstrated that ChatGPT Atlas is vulnerable to clipboard injection attacks. Atlas' agent mode might click on a malicious link that hijacks your clipboard without you knowing it, leading you to paste a malicious link in your browser.

With OpenAI’s release of the ChatGPT Atlas web browser, we evaluated its reviews next to Perplexity’s Comet in the battle that’s reshaping how we use the web.

OpenAI's new AI browser ChatGPT Atlas can mimic human clicks, potentially inflating ad costs and distorting website analytics.

Experts caution that AI-powered browsers like ChatGPT Atlas could open the door to new kinds of attacks—from prompt injections to data leaks—as artificial intelligence becomes more deeply woven into the web.

Despite safeguards, OpenAI's new AI browser 'can still make (sometimes surprising!) mistakes,' warns CISO Dane Stuckey, who also cites prompt injection attacks as an 'emerging risk.'