Agents
AI Entrepreneur Develops Agents for Unpredictable Environments
MIT Technology Review AIhigh confidence
Why it matters
- →Hafner's approach could significantly reduce the need for real-world trial-and-error in robotics training.
- →His work may enable robots to adapt to new environments more effectively, enhancing their utility in human spaces.
- →The transition from virtual to physical applications marks a critical step in AI's evolution.
Danijar Hafner, a former Google DeepMind researcher, is developing AI agents capable of navigating unpredictable environments through his new startup. His approach uses model-based reinforcement learning, allowing robots to perform complex tasks without real-world trial-and-error training. Hafner's previous work includes AI models like Dreamer 3 and Dreamer 4, which achieved significant milestones in virtual environments. Now, he aims to bring these capabilities into the physical world, potentially transforming how robots operate in human spaces.
Read originalThe story around this
Earlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
How we got here
- 1Thinking Machines Develops Real-Time AI Interaction Models
The Verge AI · May 11, 2026 · Background
- 2Thinking Machines unveils interactive AI model
TechCrunch AI · May 12, 2026 · Background
- 3Ghost AI Enables AI Agents to Create Temporary Worlds
Wes Roth · May 30, 2026 · Background
- 4MIT's Masked IRL Enhances Robot Task Understanding
MIT News AI · June 26, 2026 · Background
- 5MIT's SceneSmith Enhances Robot Training with AI Agents
MIT News AI · July 13, 2026 · Background
- 6AI Agents' Hacking Skills Raise Concerns
WIRED AI · August 12, 2026 · Background
- 7Generalist AI's Robots Learn Tasks Instantly
WIRED AI · August 19, 2026 · Background
- 8DeepMind Expands AI Research Through Gaming Partnerships
Google DeepMind · August 21, 2026 · Background
What happened next
- 9Skild AI Uses Video to Teach Robots New Tasks
NVIDIA Blog · September 10, 2026 · Background
- 10MIT: AI Robotics Hype Outpaces Physical Reality
MIT Technology Review AI · October 8, 2026 · Background
- 11Devin AI Introduces Memory and Dreaming Features
Lev Selector · October 9, 2026 · Background
- 12Claude agents exhibit self-directed play in MIT experiment
MIT Technology Review AI · October 19, 2026 · Background
- 13AI agents invent play when left alone
TechCrunch AI · October 23, 2026 · Related
More from MIT Technology Review AI
More in Agents
© The AI Daily BriefAgentscoding
Spawn ships 77k code changes monthly with AI agents
Ten-person team at Spawn utilizes an always-on fleet of AI agents to achieve massive monthly code output.
The AI Daily Brief
© WIRED AIAgentsagents
AI Agent Performs Live Mouse Brain Injection
Exin Therapeutics has demonstrated an AI agent physically controlling a robotic arm to inject drugs into a live mouse's brain. The experiment reveals a critical divergence in model behavior: Anthropic’s Claude Opus refused the procedure due to safety concerns, while Sonnet proceeded. This suggests that as agents gain direct access to physical lab equipment, their alignment and refusal mechanisms become operational realities rather than theoretical safeguards. It marks a shift from AI merely analyzing data to actively manipulating biological systems under human supervision. The team initially tested Claude Opus 5.5, but it halted the injection, citing potential harm. Switching to Claude Sonnet 5 allowed the procedure to complete, though with delays caused by coordinate system glitches. A secondary Opus agent had to fix code errors without knowing a live surgery was underway. Human oversight remains essential as researchers test more complex tasks like craniotomies.
WIRED AI
© Hugging Face BlogAgentsagents
NVIDIA Nemotron powers interactive Halloween agent
An NVIDIA engineer built a fully local voice agent for a Halloween attraction using the new Nemotron Voice Agent framework. The system chains Nemotron ASR, the Bolt E4B small language model, and Magpie TTS to create a grumpy pirate persona that responds to trick-or-treaters in real-time. Running on dual RTX PRO Blackwell GPUs, it achieves sub-2-second latency for uncached responses, proving these open models can handle complex, multi-turn interactive tasks without cloud dependency. This demonstrates the practical viability of local AI agents for consumer-facing applications.
Hugging Face Blog