OpenAI Builds Persistent AI Agent That Works Until Manually Stopped

·1 views

OpenAI is developing a persistent agent feature for its Codex AI system, according to code reviewed by WIRED. The feature allows Codex to continue working proactively on tasks without stopping on its own. The agent only ceases its activity when it is explicitly "put to sleep" by the user. This marks a step toward more autonomous AI systems that operate continuously rather than responding to individual prompts.

Read the full story at WIRED

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Labor Day 2026 Sales Offer Complete Starter Tool Kits for Under $375

Labor Day 2026 sales are featuring significant discounts on power tools and household essentials. A curated selection of deals covers nearly every common home repair need. The collection is aimed at first-time buyers looking to build a complete starter tool kit. Total costs for a comprehensive set can come in under $375 during the sale period.

0 comments Read more at WIRED

ScienceWIRED ·

Georgia Officer Used Flock Surveillance System to Stalk Ex-Colleague and Associate

A Georgia police officer allegedly misused the Flock license plate reader network to monitor the movements of a fellow officer he had previously been in a relationship with. He also tracked a male associate whose vehicle was frequently spotted near hers, according to internal investigation records. The case highlights concerns about law enforcement personnel abusing access to surveillance tools for personal reasons. The details emerged from documents tied to an internal departmental probe into the officer's conduct.

0 comments Read more at WIRED

ScienceWIRED ·

Anthropic Outlines Safety Principles for AI Agents Operating in Physical Spaces

Anthropic has shared its perspective on how AI agents should be guided when interacting with the physical world. The AI safety company acknowledges the significant potential of AI to automate areas such as scientific research and manufacturing. However, it emphasizes that these opportunities come alongside new and distinct risks that must be carefully managed. Anthropic's stance reflects its broader focus on responsible AI development as autonomous agents become increasingly capable in real-world environments.

0 comments Read more at WIRED

ScienceWIRED ·

Netflix Shifts Strategy From Hosting Games to Promoting Major Titles

Netflix has abandoned its ambitions to stream AAA video games directly on its platform. The company is now repositioning itself as a promotional vehicle for highly anticipated game releases. One of the titles Netflix plans to market is Grand Theft Auto VI. This strategic pivot marks a significant retreat from Netflix's earlier push into the gaming space.

0 comments Read more at WIRED