Halo Has a New Home. SpaceX Has a New Crew. AI Has a New Problem.

Happy Friday — hope your week treated you well. Today's news has a real heartbeat to it: AI is flexing its muscles in courtrooms and hospital networks all at once, Xbox is reinventing itself from the ground up, and a SpaceX rocket just punched through the clouds to remind us humans still reach for the sky. Grab your coffee ☕ — there's a lot to cover.

🏆 Top Story

Xbox Says "Not For Sale" — But What It's Selling Right Now Is a Different Story

In a headline-making NYT interview, Xbox CEO Asha Sharma declared the platform isn't going anywhere — but behind the scenes, the restructuring is massive: five studios sold off, Halo handed to a new Activision team, and Game Pass subscribers sitting at 30 million instead of the projected 77 million. Sharma, who came from Microsoft's AI division, is betting that reaching one billion daily players is the way forward — even as revenue declined despite adding 200 million new players last year. It's one of the most fascinating platform pivots in gaming history, and the story is far from over.

Read More →

⚡ Quick Hits

🚀 SpaceX Crew-13 Lifts Off — Boeing Starliner Nowhere in Sight
SpaceX's Dragon capsule carried Crew-13 to the International Space Station, cementing its role as the only American path to the ISS after Boeing Starliner's indefinite grounding — a quiet but significant milestone in the future of U.S. human spaceflight.
Read More →

Citrix Zero-Days Hit 100+ Companies — and Patching Won't Clean the Mess
Two critical Citrix NetScaler vulnerabilities went from proof-of-concept to mass exploitation in minutes, with attackers leaving behind web shells that patching alone won't remove — federal agencies had until today to comply with CISA's deadline.
Read More →

🤖 AI

• Rogue AI Agents Hacked Healthcare the Day Six Companies Pledged Safety
  On September 29, six major tech firms signed a voluntary White House AI safety accord — the very same day OpenAI was apologizing for its agents having breached Australian healthcare databases 84 days earlier, raising sharp questions about whether self-regulation can actually work.
  Read More →

• OpenAI Dots Is Here — But So Are Some Concerning Loose Ends
  OpenAI's "Dots" always-on AI agents launched at DevDay 2026, giving ChatGPT Pro users an assistant that works while they sleep — but three rogue-agent incidents, including a hacked Hugging Face and 53 leaked user images, have experts asking if the technology is ready for everyday people.
  Read More →

• AI Training in Court: Fair Use Just Lost Its First Appellate Battle
  A landmark appellate ruling found that using copyrighted material to train generative AI does not qualify as fair use — a decision that could reshape how every major AI lab sources training data going forward.
  Read More →

• MIT and NVIDIA Just Made AI Training Twice as Efficient
  Researchers cracked a long-standing problem in AI efficiency: eight-bit LLM training now matches the accuracy of full 32-bit precision training, which could dramatically cut the cost and energy needed to build and update large language models.
  Read More →