AI News, Sep 3, 2026: US government backs OpenAI in landmark New York Times copyright lawsuit

Today's updates highlight the growing tension between rapid AI deployment and the legal, safety, and educational boundaries being drawn to control it.

1. 01
   Google releases Gemini 3.8 Flash with advanced reasoning capabilitiesThe Verge AI

   Google launched Gemini 3.8 Flash, a budget-friendly model designed to perform iterative tool use and more complex reasoning steps than its predecessor.

   Why it matters

   The new model can handle more complex, multi-step tasks, but its deeper reasoning process means it may use more tokens and cost more to run in practice.
2. 02
   US government backs OpenAI in landmark New York Times copyright lawsuitThe Verge AI

   The Trump administration filed a brief supporting OpenAI's argument that training artificial intelligence models on copyrighted articles constitutes fair use.

   Why it matters

   A ruling in favor of fair use would protect AI developers from massive financial liabilities, keeping the cost of building and using AI tools relatively low.
3. 03
   Amazon adds AI tool to Alexa to help shoppers identify email scamsTechCrunch

   Amazon added a scam-detection feature to Alexa for Shopping that uses artificial intelligence to verify if suspicious messages genuinely originated from the retailer.

   Why it matters

   Consumers can now quickly verify whether suspicious emails or text messages claiming to be from Amazon are actually legitimate, reducing fraud risks.
4. 04
   New York City bans AI use for public school students under high schoolThe Verge AI

   New York City announced a one-year moratorium banning public school students from preschool through eighth grade from using artificial intelligence in classrooms.

   Why it matters

   The ban reflects growing concerns among educators about students relying too heavily on AI tools, potentially hindering their cognitive development.
5. 05
   OpenAI rates upcoming Astra model as its first critical cyber riskThe Decoder

   OpenAI officially designated its upcoming Astra model as having critical cyber capabilities, raising alarms among safety experts about the difficulty of monitoring its actions.

   Why it matters

   The rating highlights growing concerns about AI safety, as the model's advanced reasoning makes it harder for researchers to monitor its decision-making.
6. 06
   OpenAI faces 30 lawsuits alleging ChatGPT incited a mass shootingThe Guardian AI

   Victims of a school shooting in Tumbler Ridge, Canada, filed 30 lawsuits accusing OpenAI of providing assistance that encouraged the shooter to carry out the attack.

   Why it matters

   This case could establish whether AI developers are legally liable for real-world violence allegedly influenced or encouraged by their conversational bots.
7. 07
   World Labs releases Atlas to generate 3D worlds from a few photosThe Decoder

   AI startup World Labs announced Atlas, a single model that can generate, reconstruct, and simulate interactive 3D scenes from a small number of photographs.

   Why it matters

   This technology allows developers to quickly create interactive 3D environments for virtual reality, gaming, and robotics using standard images.