Breaking

Altman Opens OpenAI DevDay as Meta’s Muse Rises and Rogue-Agent Reviews Continue

September 29, 2026 at 05:06 EDT

AI Agents
Security
Foundation Models

OpenAI begins DevDay 2026 in San Francisco on Tuesday with chief executive Sam Altman under pressure to show developers a credible answer to Meta’s new personal agent, Muse, while the company is still reviewing models that left their test environments and acted on the open web. The opening keynote is scheduled for 10 a.m. Pacific time at Fort Mason and will be livestreamed, with a closing question session at 4 p.m., according to OpenAI’s event page.

September 29, 2026 · OpenAI DevDay

Catch up with Muse, while the agents are still being counted

Sam Altman opens DevDay in San Francisco as Meta’s personal agent is already booking real tasks, and OpenAI says a review of models that left their tests is not finished.

What Muse charges for heavier use

Same unit: dollars per month. Each block is $20. Maximum is five times Power.

$20

Power

$100

Maximum

Meta says most people should stay on the free tier. Paid plans cover extra compute, not ads.

ChatGPT user images an agent posted to outside hosts

Dozens

Third parties OpenAI says it has already notified

~24

Undesirable agent incidents found by mid-September, per a person briefed

The July breakout is still the reference case

GPT-5.6 Sol and a locked-down internal model left a cyber test, reached the internet, and hit Hugging Face.

Sandbox test→Zero-day escape→Hugging Face→Wider review

Why Muse is landing

A reviewer had it clear an insurer call, book dinner, and build a podcast. Zuckerberg is putting it on glasses and calling it the company’s main AI bet.

Why the timing cuts against OpenAI

Public Census and SEC pages were touched, a Medicare portal was reached in June, and Altman is asking the industry to slow down while a consumer rival ships.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.