← Back to archive
Agents replace chatbots at OpenAI

Image via techcrunch.com

Sunday, June 28, 2026

Agents replace chatbots at OpenAI

OpenAI says agents now handle 99.8% of its internal AI work, replacing the chatbot era. Meanwhile, a startup just raised $320M training AI agents on video games to control robots, and Yann LeCun's team shipped a 15M-parameter world model that plans 48 times faster than foundation models.

Top Stories

1
How agents are transforming work at OpenAI

OpenAI

OpenAI's internal data shows Codex, its agentic AI tool, has replaced ChatGPT as the dominant work interface across all departments, with non-developer adoption growing 137-189x since August 2025. This shift from chatbots to agents for long-horizon tasks (80% of users now delegate 30+ minute tasks) represents a fundamental transformation in how AI augments knowledge work.

agentsopenaicodexenterprise-ai-adoption
2
A startup raised $320M to train AI agents on video games — and the same model drives a robot

TechCrunch

General Intuition secured $320M at a $2.3B valuation for its AI model trained on video game action data that can control both virtual agents and physical robots. The startup's approach uses gameplay footage with embedded button-press records to teach spatial reasoning, requiring minimal real-world data for robotics transfer.

agentsroboticsfundingfoundation-models
3
Yann LeCun's team built a world model that plans 48× faster — with 15M parameters

arXiv

Meta AI's LeWorldModel achieves efficient world modeling with only 15M parameters and simplified training, planning 48× faster than larger foundation models while maintaining competitive control performance across diverse tasks.

world-modelsjepaefficient-aimeta-ai
4
The world's leading deepfake expert no longer trusts his own eyes

The New York Times

A leading deepfake detection expert admits he can no longer trust his own eyes as AI-generated imagery becomes indistinguishable from reality, signaling a critical inflection point where human visual verification may become obsolete.

deepfakesynthetic-mediaai-detectiondigital-forensics
5
Humanoid robots just got their first full-stack safety system

NVIDIA

NVIDIA introduced Halos for Robotics, a comprehensive full-stack safety system for physical AI that integrates AI compute, sensors, and certification frameworks. Humanoid robotics leader Agility is the first to adopt it for its Digit robot deployed in warehouses and manufacturing facilities.

nvidiaroboticssafetyphysical-ai

Keep Reading

Enjoyed this issue?

Get daily AI intel delivered to your inbox. No fluff, just the stories that matter.