Speak About AI
Back to Resources

This Week In AI - Feb 10, 2026

By Robert Strong•Feb 10, 2026
The week the robots finally decided to start talking back (and they’re surprisingly sassy).

Speak About AI has sifted through the week’s deep-tech musings, the visionary predictions, and the billionaire spats to bring you the up-to-date "vibes" of the AI industry. Here’s what the heavy hitters were buzzing about this past week.

1. The Great "Ad-Off": OpenAI vs. Anthropic

Sam Altman (OpenAI CEO) and the folks at Anthropic basically spent their week taking shots at each other over OpenAI’s planned introduction of ads.

  • The Drama: Anthropic launched a series of "Super Bowl-adjacent" ads ads poking fun at the idea of targeted ads appearing in chatbots. Their slogan? "Ads are coming to AI. But not to Claude".
  • Sam’s Clapback: Sam Altman wasn't having it. He called the ads called the ads "clearly dishonest," while simultaneously defending OpenAI's move to include ads as a way to keep AI "accessible" and "free" for everyone.
  • The Analysis: Whether you want a six-pack or dating advice, OpenAI thinks you deserve a free chatbot with a side of commercial breaks. Anthropic is on the attack because they haven’t come to the same conclusion regarding their free-tier chatbot service (yet).

2. Andrew Ng and the Rise of "Sovereign AI"

The "Teacher of AI" himself, Andrew Ng, was out here warning us about a breakup—between the U.S. and its allies.

  • The Take: Andrew argued that U.S. policies (like export controls and "America First" rhetoric) are accidentally pushing other countries to build their own "Sovereign AI". Essentially, Andrew is saying, "If you don't share your toys, everyone is just going to go build their own sandbox".
  • The Solution: He’s rooting for open-source models like DeepSeek and Qwen, arguing that nations don't need to build everything from scratch; they just need to use open-weight models to ensure they aren't at the mercy of a single foreign power.

3. Demis Hassabis: 100-Hour Work Weeks & AGI Timelines

Google DeepMind's CEO, Demis Hassabis, gave us a glimpse into his life, and frankly, I need a nap just reading about it.

  • The Grind: Demis revealed he works 100-hour weeks. (And here I am, feeling proud when I remember to clear my inbox.)
  • The Prediction: He’s still betting on 2030 for AGI (Artificial General Intelligence), but he sets a high bar: it has to be capable of "scientific creativity" (e.g., setting hypotheses rather than just testing them) and continuous learning.
  • The Vibe: He’s ignoring the "noise" of the competition and focusing on multimodal AI—the kind of tech that will power smart glasses that replace smartphones and robots that actually understand the world.

4. Allie K. Miller: From Prompting to Verifying

Allie K. Miller, our go-to strategist, dropped some truth bombs for anyone feeling overwhelmed.

  • The Shift: She predicts that in 2026, you won’t even be "prompting" AI anymore—the AI will be prompting you.
  • The Skill: Your new job? Verification. The most important skill isn’t knowing how to write a prompt; it’s knowing if the AI’s output is actually good or just a very confident hallucination.
  • The Methodology: She’s pushing her "3P methodology"—People, Process, and Product—urging companies to treat AI like a teammate, not just a tool.

5. Quick Hits from the Front Lines

  • Yann LeCun: The Meta legend has launched AMI Labs in Paris, focusing on building "world models" that understand physics rather than just predicting the next word in a text.
  • The Market: Microsoft had a rough week, losing $381 billion in market value because investors are suddenly realizing that building $100 billion AI data centers is, well, expensive.
  • Deepfakes: A sobering reminder from Hany Farid (UC Berkeley) that deepfakes are no longer "novel"—they are now "routine, scalable, and cheap," making "seeing is believing" a thing of the past.