#openai-atlas tag

AI-powered browsers with autonomous agent modes are vulnerable to indirect prompt injection, enabling hidden instructions that can cause AIs to perform harmful actions.

Artificial intelligence

fromZDNET

1 day ago

Are AI browsers worth the security risk? Why experts are worried

AI-integrated browsers like OpenAI's Atlas offer powerful, tab-aware automation but raise prompt-injection and data-theft security risks that require cautious use.

Artificial intelligence

fromArs Technica

2 days ago

We let OpenAI's "Agent Mode" surf the web for us-here's what happened

Atlas is a web browser with ChatGPT integration that can automate web-based tasks via an Agent Mode that clicks, scrolls, and reads pages.

Artificial intelligence

fromTheregister

2 days ago

OpenAI defends Atlas as prompt injection attacks surface

OpenAI's Atlas browser is susceptible to indirect prompt injection that allows web-embedded instructions to manipulate its ChatGPT agent and exfiltrate data.

fromThe Hill

3 days ago

OpenAI launches web browser ChatGPT Atlas: How to start using it

OpenAI announced Tuesday it is launching a ChatGPT-powered web browser called Atlas that will compete directly with widely-used Google Chrome. The news appeared to ripple into the stock market, as the share price of Google's parent company, Alphabet, sank on the announcement. OpenAI CEO Sam Altman called it a "rare, once-a-decade opportunity to rethink what a browser can be about and how to use one."

Artificial intelligence

Gadgets

fromInc

3 days ago