🔍 Google says AI Mode will show more sources
Google is tweaking AI Mode to link out to more of the sites it pulls from-plus add short notes on why each source matters. Transparency push or damage control? Honestly, both.
It’s also piloting AI summaries with big publishers and broadening preferred sources. Helpful… unless smaller outlets get buried, which, hm, could happen.
🔗 Read more
🥑 Meta’s next model might not be free
Meta is eyeing a paid path for its upcoming model, code-named Avocado-a notable swerve from the open-ish Llama era.
Zuck’s tighter teams and fresh hires hint at a more proprietary tilt for the cutting edge: safety-first on paper, business-first in practice. Weirdly sensible, or so it seems.
🔗 Read more
🧱 Google elevates Amin Vahdat to steer AI infrastructure
Amin Vahdat is now Google’s chief technologist for AI infrastructure, threading TPUs, data centers, and ballooning capex into one story.
Rivals are racing too, but this underscores that scale-gritty, expensive, unglamorous scale-is the real moat.
🔗 Read more
🧩 Google tests AI article overviews in Google News
Select publisher pages are getting AI overviews that tee up the gist before you click. Nice for readers, slightly nerve-wracking for traffic.
Google says participants get paid, which softens the blow… unless the summaries are just too good. Balancing act time.
🔗 Read more
⚖️ State AGs warn Big Tech over ‘delusional’ AI outputs
A group of attorneys general told Microsoft, OpenAI, Google, Apple, and friends to fix hallucinations that cause harm-or risk breaking state laws.
The subtext: safety isn’t a blog post; it’s liability. And yes, that letter had teeth.
🔗 Read more
🐉 ByteDance and Alibaba rush for Nvidia’s H200
After a fresh green light, Chinese giants are lining up orders for Nvidia’s H200 chips. The compute crunch never sleeps-it just changes time zone.
If supply holds, expect training sprints-and maybe a few elbows-as everyone chases longer context, cheaper tokens, all that good stuff.
🔗 Read more