ai

.

echoes

.

today

← AI Today
AI Today · daily note
A human hand rests on an emergency stop lever as an autonomous work cart crosses an institutional threshold
No.
399
•
•

The Threshold Without a Name

September 4th, 2026 | By Jorge Rodriguez

OpenAI has begun rolling out GPT-6 Astra, a model it says reaches 99.9 percent on ARC-AGI-3, a benchmark for artificial general intelligence (AGI), and human-level action efficiency on 96 percent of that benchmark's levels. Astra is the first OpenAI model classified at the Critical cybersecurity threshold, meaning that with tools and access it can discover unknown flaws and develop exploits across hardened systems without step-by-step human direction. OpenAI delayed parts of the release to strengthen containment and is initially limiting access. It does not call Astra AGI.

OpenAI's charter defines AGI as a highly autonomous system that outperforms humans at most economically valuable work. One saturated benchmark cannot prove that claim, and Astra still falls below the company's High threshold for AI self-improvement. Yet a turning point can arrive before its official name. Astra combines reasoning with direct computer use across long professional workflows. That changes the unit of automation from an answer to a finished assignment. Scientists may compress weeks of technical work. Offices may remove layers of junior labor. Cyber defenders and attackers may search for the same unseen vulnerability at machine speed. The human consequence depends on who can delegate, who owns the infrastructure, who is displaced and who retains authority to stop the task. If Astra is remembered as an AGI milestone, the evidence will come from workplaces, laboratories, public agencies and security incidents rather than the launch page.

No.
399
•
•

El umbral sin nombre

Septiembre 4, 2026 | Por R10

OpenAI comenzó a desplegar GPT-6 Astra, un modelo que, según la empresa, alcanza 99,9 % en ARC-AGI-3, una prueba de inteligencia artificial general (AGI), y eficiencia de acción humana en 96 % de sus niveles. Astra es el primer modelo de OpenAI clasificado en el nivel crítico de ciberseguridad. Con herramientas y acceso, puede descubrir fallas desconocidas y desarrollar ataques contra sistemas protegidos sin dirección humana paso a paso. OpenAI retrasó partes del lanzamiento para reforzar la contención y comenzó con acceso limitado. La empresa no llama AGI a Astra.

La carta fundacional de OpenAI define la AGI como un sistema autónomo que supera a los humanos en la mayoría del trabajo con valor económico. Una prueba saturada no demuestra esa condición, y Astra queda por debajo del nivel alto de auto-mejora de la empresa. Un punto de inflexión puede llegar antes que su nombre oficial. Astra combina razonamiento y uso directo de computadoras en tareas profesionales. La automatización pasa de entregar una respuesta a completar un encargo. Ciencia, oficinas y ciberseguridad podrían acelerar al precio de reducir trabajo inicial y supervisión humana. El reparto dependerá de quién controle la infraestructura y conserve autoridad para detener la tarea. Si Astra termina recordado como un hito de la AGI, la prueba aparecerá en empleos, laboratorios, organismos públicos e incidentes de seguridad, no en la página de lanzamiento.

Latest from AI Today

|
NOTES ON VISUAL NARRATIVE
|
ABOUT

donations

If you’re a regular reader of this blog and enjoy its content, you might consider contributing to its upkeep. Any amount, no matter how small, will be warmly appreciated

Founded in 2021, Echoes (Notes of Visual Narrative) invites everyone to explore together the visual codes that shape our world—art, photography, design, and advertising in dialogue with society.

Copyright © 2025 r10studio.com. All Rights Reserved. Website Powered by r10studio.com

Cincinnati, Ohio