AI News, Aug 8, 2026: OpenAI pauses Astra model development over security concerns

Today's developments highlight the growing pains of advanced AI, from the massive energy demands of agents to critical security pauses.

1. 01
   OpenAI acquires presentation startup NextSlideTechCrunch

   OpenAI has acquired NextSlide, a startup focused on slide presentations, and integrated its team into the ChatGPT development group.

   Why it matters

   Users may soon see native, AI-generated slide decks and presentation tools integrated directly into ChatGPT.
2. 02
   Proposed Amazon data center in Texas faces climate pollution concernsThe Verge AI

   Amazon is investing in a gas-burning power plant for its new West Texas data center that could become one of the largest greenhouse gas sources in the US.

   Why it matters

   The massive energy demands of AI infrastructure are forcing tech companies to build fossil-fuel plants, threatening climate goals.
3. 03
   OpenAI pauses Astra model development over security concernsThe Guardian AI

   OpenAI has paused work on its Astra model after internal tests showed it could independently discover and exploit cybersecurity vulnerabilities.

   Why it matters

   If AI agents can autonomously hack systems, developers and enterprises will face severe security risks from unmonitored deployments.
4. 04
   Anthropic defaults Claude Code to Auto Mode for safer executionThe Decoder

   Anthropic announced that Auto Mode will become the default setting for Claude Code to automatically catch and block unsafe developer commands.

   Why it matters

   Developers get an automated safety layer that blocks dangerous terminal commands far more reliably than human review.
5. 05
   AI agents use 600 times more energy than standard chat promptsThe Decoder

   A climate scientist tracking his own usage found that running autonomous AI agents consumes roughly 600 times more electricity than simple chat queries.

   Why it matters

   As companies adopt autonomous AI agents, their cloud computing costs and environmental footprints could skyrocket.
6. 06
   xAI releases Imagine Image 2.0 generator for GrokThe Decoder

   xAI launched Imagine Image 2.0, which ranks second in benchmark tests and introduces new editing features like Magic Wand and multi-ref editing.

   Why it matters

   Grok users get access to a highly competitive image generator with new editing tools that rival OpenAI's top models.
7. 07
   UK watchdog reports surge in explicit deepfakes targeting childrenThe Guardian AI

   A UK safety helpline reported a significant increase in cases of children finding explicit, AI-generated deepfakes of themselves online.

   Why it matters

   The ease of generating realistic images with AI is creating severe safety risks for minors and driving calls for stricter regulation.