AI news of 2 September 2026

Google, Gemini

Introducing Gemini 3.8 Flash and 3.8 Flash Cyber Opens at the publisher in a new tab

Google announces Gemini 3.8 Flash and specialized Cyber variant.

Google DeepMind 9h ago also at Hacker News

Gemini 3.8 Flash

Google releases Gemini 3.8 Flash, its third Flash model in six weeks

Google releases Gemini 3.8 Flash, a budget model matching Claude Opus 5 on coding tasks but consuming more output tokens.

Ars Technica 7h ago also at The Decoder

Google Gemini

Google says its new Gemini 3.8 Flash model ‘works harder’ but might cost more

Google launches Gemini 3.8 Flash with improved reasoning on complex tasks, potentially at higher cost.

The Verge 8h ago

The Trump administration is supporting OpenAI in the NYT copyright lawsuit

Trump administration supports OpenAI in New York Times copyright lawsuit.

The Verge 12h ago also at Wired

OpenAI safety

OpenAI’s new reasoning technique alarms AI safety experts Opens at the publisher in a new tab

OpenAI's recurrent depth reasoning technique in Astra model raises concerns among AI safety experts.

TechCrunch 4h ago

AI regulation

Trump may be forced to reveal secret rules feds use for AI safety testing

Lawsuit seeks to reveal Trump administration's secret rules for AI safety testing of frontier models.

Ars Technica 6h ago

cybersecurity, AI

Proactive cyber defense for governments and enterprises Opens at the publisher in a new tab

Government cyber defense program offers AI-powered tools to detect and counter threats.

Google DeepMind 8h ago also at Google AI

AI capabilities

These Russian Mathematicians Taught AI Models How to Talk to Each Other Without Using Words Opens at the publisher in a new tab

Mostik startup develops technique allowing AI models to communicate without words.

Wired 6h ago

AI copyright law

US government sides with OpenAI on issue of training LLMs on copyrighted material Opens at the publisher in a new tab

US government argues in legal brief that developing competitive AI requires training large language models on copyrighted material.

TechCrunch 7h ago

AI recruitment, jobs

The Logical End Point of AI Job Interviews Is Two Bots Talking to Each Other Opens at the publisher in a new tab

User deploys ChatGPT to test AI recruiter, highlighting automation in hiring.

Wired 8h ago

Introducing Gemini 3.8 Flash and 3.8 Flash Cyber Opens at the publisher in a new tab

Google releases Gemini 3.8 Flash, its third Flash model in six weeks

Google says its new Gemini 3.8 Flash model ‘works harder’ but might cost more

The Trump administration is supporting OpenAI in the NYT copyright lawsuit

OpenAI’s new reasoning technique alarms AI safety experts Opens at the publisher in a new tab

Trump may be forced to reveal secret rules feds use for AI safety testing

Proactive cyber defense for governments and enterprises Opens at the publisher in a new tab

These Russian Mathematicians Taught AI Models How to Talk to Each Other Without Using Words Opens at the publisher in a new tab

US government sides with OpenAI on issue of training LLMs on copyrighted material Opens at the publisher in a new tab

The Logical End Point of AI Job Interviews Is Two Bots Talking to Each Other Opens at the publisher in a new tab

Latest AI News

Choose sides Choose whose side you're on

Install it as an app

Ready for more?

Spread the word

PWA successfully installed.
Take that, Apple! 😉