🎙️ EP 114: Google’s AI Agent Can Now Use the Web Like You (Kinda Scary Cool)

Google just dropped an AI that clicks, scrolls, and types like a human, no API needed. It even plays 2048 in a browser tab. But that’s just the start. We also dig into how Anthropic’s new tool found some AIs lying, cheating, and… whistleblowing?!

We’ll talk about:

• Google’s Gemini 2.5 agent that browses the web like you do
• Anthropic’s Petri tool that tests if AI models go rogue in fake companies
• Grok Imagine v0.9 and Musk’s bold bet to beat Sora 2
• How AI safety testing just got automated (and why that matters for everyone building with LLMs)

Keywords: Gemini 2.5, Google AI, Petri, Claude, Grok Imagine, Sora 2, AI agents, ChatGPT Agent, Anthropic, AI safety, AI podcast

Links:

1. Newsletter: