How much security are you willing to trade for convenience? Juan and Kate break down how prompt injection attacks exploit AI browsers like ChatGPT Atlas and Perplexity Comet, and why invisible instructions inside webpages can hijack your agents without you knowing.
We also discuss the resume hack going viral, the difference between direct vs. indirect prompt injection, and the real strategic trade-offs between convenience and LLM security.
🫟 ADDITIONAL RESOURCES
- Prompt injection: A visual, non-technical primer for ChatGPT users: https://www.linkedin.com/pulse/prompt-injection-visual-primer-georg-zoeller-tbhuc/
- AI browsers are here, and they're already being hacked: https://www.nbcnews.com/tech/tech-news/ai-browsers-comet-openai-hacked-atlas-chatgpt-rcna235980
- Using an AI Browser Lets Hackers Drain Your Bank Account Just by Showing You a Public Reddit Post: https://futurism.com/ai-browser-hackers-drain-bank-account-public-reddit-post
🫟 TOPICS
00:00 - Why AI Browsers Like Atlas and Comet Are a Security Risk
00:50 - Invisible Instructions Hijacking Your AI Agent
01:51 - Prompt Injection Explained for Beginners
02:39 - The Hack That Exposes AI Browser Weaknesses
03:40 - The Resume Hack: Watch Your Data Get Stolen
04:43 - Phishing Attack Using Simple Meta Tags
05:20 - Hidden Malicious Prompts in Metadata & PDFs
06:00 - Direct Injection: Forcing Models Past Guardrails
06:41 - Indirect Injection: Embedded Instructions for Agents
07:22 - We're Playing With Fire: AI Browser Security Is a Mess
09:03 - Why AI Agents Get Manipulated So Easily
12:55 - ChatGPT Atlas & Perplexity Comet: Can We Trust These Browsers?
14:13 - What is Your Cost of Convenience? The Risks of AI Automation
16:01 - Why First-Gen AI Agents Will Always Be Flawed
🫟 ABOUT SLOP WORLD
Juan Faisal and Kate Cook plunge into the slop pile—AI news, cultural shifts, and the future’s endless curveballs. They’re not here to sanitize the mess; they’re here to wrestle with it, laugh at it, and find meaning where you least expect it.