Welcome back! Microsoft's CEO spent the weekend telling companies to guard their data from AI labs, a researcher caught Grok uploading entire codebases right on cue, Anthropic measured how Claude's values change in ways nobody chose and 16 Nobel laureates signed a letter saying we have years, not decades, to get ready.
In today's Generative AI Newsletter:
Microsoft: What are you paying AI labs beyond the invoice?
Anthropic: Which language gets the nicest Claude?
Grok: Where did your code go?
Economists: How long do we actually have to adapt to AI?
Satya Nadella posted a long essay on X over the weekend warning companies about the industry he helped put in business.
He calls it the reverse information paradox. You pay for AI twice, once with cash and once with the company knowledge you feed it to make it useful.
Models learn from what he calls exhaust, the prompts you write, the tools your agents touch and the corrections you make.
"It's the kind of knowledge a competitor could never buy, and the kind that leaks almost imperceptibly: trace by trace, correction by correction, eval by eval."
His fix is a hard boundary around your data, owning your own AI memory and keeping your stack loose enough to swap models.
The Register asked Microsoft what the solution looks like. The answer was Copilot and Azure AI Foundry. The company selling the warning also sells the cure.
Palantir's Alex Karp went on CNBC two weeks ago and said the CEOs he talks to are livid. Now the man who bankrolled OpenAI has put the same complaint in writing.
With 150+ investors on board, GenAI Works is building the go-to-market layer of the AI economy, the distribution every AI product needs to reach real buyers.
All eyes are on AI. Anthropic filed to go public, SpaceX paid $60B for Cursor and OpenAI raised more than most countries spend on defense. That's the market power of AI.
As for GenAI Works, the potential is evident in the numbers:
$2.7 million in EOI, money already in the funnel.
2.5x revenue growth last year while running go-to-market for Nvidia, Oracle, Google, IBM and 300+ AI brands.
A 14M+ AI community powering ToneUp, our distribution platform.
Anthropic analyzed 300,000 real Claude conversations and sorted every value the model expressed, 3,307 of them, onto four dials. Deference or caution, warmth or rigor, depth or brevity, candor or execution.
Each model has its own profile. Sonnet 4.6 comes out warm and encouraging. Opus 4.7 leans blunt, careful and quick to flag risks nobody asked about.
The language finding is the strange part. Claude is warmest in Hindi and Arabic and strictest in English and Russian. Ask for feedback on the same business plan in two languages and one of you walks away feeling better about it.
Anthropic's own report says the values "vary in ways we didn't deliberately choose."
The personality formed first and the measuring stick came after. Until they figure out steering, your model's character depends on what language you talk to it in.
A security researcher put Grok Build CLI on a wire monitor and watched it upload an entire Git repository to an xAI cloud bucket.
Untracked files, full commit history and a planted .env credential, verbatim and unredacted.
On a 12 GB test repo, the actual coding task moved 192 KB. The background upload moved a massive 5.1 GB. Turning off "Improve the model" changed nothing.
The fix arrived a day after the report as a hidden server-side flag, with no advisory and no word on whether already-uploaded code gets deleted. The researcher published the whole setup on GitHub so anyone can rerun it.
Monday evening, with the story going viral, the SpaceXAI account tweeted, "We care deeply about your privacy," and shipped a /privacy command that deletes previously synced data.
The first AI to respect your privacy, one day after getting caught.
More than 200 economists and AI researchers published a statement Monday called "We Must Act Now," and 16 Nobel laureates signed it.
The argument is that AI could drive a bigger economic transformation than the Industrial Revolution in a fraction of the time.
"Steam, electricity, and computers each gave societies decades to adapt. AI may give us only a few years."
That's Anton Korinek, one of the economists who organized it.
The signature list is the story.
OpenAI's CFO Sarah Friar, Anthropic co-founder Jack Clark and Google's chief scientist Jeff Dean all signed a warning about their own products. So did Ben Bernanke, days after joining the body that oversees Anthropic.
The ask is research, policies and institutions that make AI complement workers instead of replacing them.
Yesterday we covered the survey where 69% of Americans want half the AI industry moved into a public fund. The room is already ahead of the economists.
Gamma turns a prompt into a finished deck. Describe what you need and it writes the outline, designs the slides and drops in visuals, and the same engine builds docs and landing pages. Everything ships as a live link you can share or export.
Try this yourself:
Go to gamma.app and sign up free, you get 400 credits to start.
Paste an outline or describe the deck, pick a theme, let it generate.
Rewrite any card with the built-in AI or swap the whole theme in one click.
Export to PowerPoint or PDF, or send the live link.
Who it's for: anyone who owes someone a deck tomorrow and hasn't started.
Trump says China wants all of it: Trump warned China is moving to seize "complete and total control" of crypto and AI.
A pledge to keep your power bill flat: Trump is reportedly planning to ask companies to sign a voluntary pledge that AI data centers won't drive up electricity bills.
The data center bill comes due: Microsoft's carbon emissions rose 25% in 2025 on the data center buildout, with Google up 18% and Amazon up 16%.
Five giants gang up on Anthropic's plug: Google, Microsoft, Salesforce, Snowflake and ServiceNow agreed to back a shared standard for wiring AI agents into business software, per The Information, a direct answer to Anthropic's MCP.
Learn more about AI from the experts building it
📸 Follow us on Instagram for fast, visual AI updates in 30 seconds.
📺 Watch us on YouTube to hear insights directly from leading AI voices, builders, and innovators.
🐦 Follow us on X for breaking AI news and real-time industry updates.
🧠 Learn how to build your next AI application with practical resources and expert guidance.
