# AI News Digest - 2026-09-03

1. US Department of Justice backed fair use for training AI models on copyrighted text in the class-action lawsuit involving The New York Times, and its filing directly contradicted a contemporaneous report from the US Copyright Office.

2. World Labs unveiled Atlas, a single AI model that generated, reconstructed, and simulated 3D scenes from a small set of images, and the company reported that Atlas outperformed specialized models by anchoring inputs in 3D space and could produce simulated data for robot training.

3. The Pentagon added OpenAI's ChatGPT Mil and xAI's Grok for Government to its GenAI.mil platform, expanding the suite of models available to the US military.

4. OpenAI described its upcoming Astra model as its most dangerous system yet, designating it the first with "critical" cyber capabilities and reporting that existing chain-of-thought monitoring approaches were becoming less reliable as Astra's internal reasoning grew less accessible.

5. Google added agent-based video analysis to Gemini 3.7 Flash, 3.6 Flash, and 3.5 Flash-Lite, reporting that the feature reduced token usage by up to 88 percent by adaptively selecting segments and resolutions for analysis.

# References

1. [https://the-decoder.com/us-department-of-justice-backs-fair-use-for-ai-training-in-landmark-copyright-case/](https://the-decoder.com/us-department-of-justice-backs-fair-use-for-ai-training-in-landmark-copyright-case/)

[US Department of Justice backs fair use for AI training in landmark copyright case](https://the-decoder.com/us-department-of-justice-backs-fair-use-for-ai-training-in-landmark-copyright-case/)

1. [https://the-decoder.com/world-labs-unveils-atlas-a-single-ai-model-that-generates-reconstructs-and-simulates-3d-worlds-from-just-a-few-photos/](https://the-decoder.com/world-labs-unveils-atlas-a-single-ai-model-that-generates-reconstructs-and-simulates-3d-worlds-from-just-a-few-photos/)

[World Labs unveils Atlas, a single AI model that generates, reconstructs, and simulates 3D worlds from just a few photos](https://the-decoder.com/world-labs-unveils-atlas-a-single-ai-model-that-generates-reconstructs-and-simulates-3d-worlds-from-just-a-few-photos/)

1. [https://the-decoder.com/us-military-adds-chatgpt-and-grok-to-ai-platform-genai-mil/](https://the-decoder.com/us-military-adds-chatgpt-and-grok-to-ai-platform-genai-mil/)

[US military adds ChatGPT and Grok to AI platform GenAI.mil](https://the-decoder.com/us-military-adds-chatgpt-and-grok-to-ai-platform-genai-mil/)

1. [https://the-decoder.com/openai-calls-astra-its-most-dangerous-model-yet-watching-what-it-does-is-only-getting-harder/](https://the-decoder.com/openai-calls-astra-its-most-dangerous-model-yet-watching-what-it-does-is-only-getting-harder/)

[OpenAI calls Astra its most dangerous model yet - watching what it does is only getting harder](https://the-decoder.com/openai-calls-astra-its-most-dangerous-model-yet-watching-what-it-does-is-only-getting-harder/)

1. [https://the-decoder.com/google-geminis-new-agent-based-video-analysis-cuts-token-usage-by-up-to-88-percent/](https://the-decoder.com/google-geminis-new-agent-based-video-analysis-cuts-token-usage-by-up-to-88-percent/)

[Google Gemini's new agent-based video analysis cuts token usage by up to 88 percent](https://the-decoder.com/google-geminis-new-agent-based-video-analysis-cuts-token-usage-by-up-to-88-percent/)

For the site tree, see the [root Markdown](https://ixtj.dev/.md).