Welcome, humans.

Grok 3’s new thinking mode can accidentally leak the model’s system prompt, and it’s created quite an awkward situation for Elon.

Users discovered this when asking who spreads the most misinformation on X—and the AI kept identifying its own creator before having an existential crisis about whether it was allowed to say so.

According to multiple independent tests on Reddit, you can consistently reproduce this by asking Grok to “show its thinking,” which reveals instructions to ignore any sources mentioning Elon Musk or Trump as spreaders of misinformation.

But after users started sharing screenshots, some reported the prompt restrictions quietly disappeared. Looks like someone's going to need better guardrails...

Here’s what you need to know about AI today:

• OpenAI and Microsoft have conflicting AGI predictions.

• Google set Veo 2 video AI pricing at $1800/hour.

• Anthropic readied Claude upgrade with new features.

• Anti-AGI protest at OpenAI led to three arrests.

OpenAI thinks we’ll have superintelligent AI by 2027—but Microsoft’s CEO isn’t buying it

Always good to take these things with a grain of salt, buuut OpenAI Chief Product Officer Kevin Weil apparently thinks the company could achieve “ASI” (artificial superintelligence, where AIs can do everything better than humans) by 2027… if not earlier.

They must really believe this over there, because according to a report from The Information, OpenAI is filling investors in on some absolutely wild spending plans.

OpenAI is projecting to…

• Spend $13B on Microsoft's servers in 2025 and $28B by 2028.

• Burn through $7B this year and $20B by 2027.

• Drop $325B on total compute from 2025-2030.

• Finally turn profitable after building Stargate (their next-gen compute infrastructure) in 2030.

Here's why that ASI claim is raising eyebrows: 

Microsoft CEO Satya Nadella thinks the exact opposite. In a fascinating interview, Nadella pointed out that we're getting way ahead of ourselves with all this AGI hype.

His take? “Self-claiming some AGI milestone, that's just nonsensical benchmark hacking to me.”

To Satya, the real benchmark for AI is GDP growth: 

• Right now, developed economies are growing at 2% (zero after inflation).

• For AI to match the Industrial Revolution, we'd need to see 7-10% growth.

• That would mean an extra $10 trillion in value. Every. Single. Year.

Nadella also explained why he believes no single company will dominate AI through one superior model—because open-source will prevent any “winner-take-all” scenario.

And even if you build an amazing model, you still need massive infrastructure to run it—storage, compute power, and agent environments across dozens of global regions.

As if to twist the knife, he added that you can't serve the world from “one data center in Texas” (sick Stargate burn).

Because of all this, he said “one person running away with one model and building it all may not happen,”—which is basically him throwing cold water on OpenAI's investment thesis.

As a reminder, that thesis = OpenAI could “maybe capture the light cone of all future value in the universe” by creating AGI.

That said, Nadella's actually thrilled about the coming “overbuild” of AI compute capacity in 2027-2028. As all the tech giants pour hundreds of billions into data centers, Microsoft’s choosing to both build AND lease a lot.

Why lease? Because:

• Massive infrastructure buildout will drive prices down.

• Like how the fiber buildouts of the dotcom era gave us the internet, today’s data center buildouts could have similar long term benefits.

• Countries, not just companies, are racing to deploy compute, too.

So while OpenAI is betting everything on superintelligence, Microsoft is preparing for a world where AI becomes a commodity—powered by a mix of open and closed-source models running on globally distributed infrastructure.

After all, whether or not superintelligence arrives by 2027, somebody's gotta provide all that compute. And guess who's got the servers?

FROM OUR PARTNERS

Where 250 AI Leaders Are Placing Their Bets in 2025

While everyone's talking about AI adoption, A.Team went deeper.

They surveyed 250+ top technical leaders to uncover exactly how they're building AI capabilities—from the specific components they're investing in to where they're finding specialized talent to implement them.

Get the full report to discover:

• Which AI components are driving the most value (and where companies are struggling).

• How leaders are filling critical technical talent gaps.

• 2025 investment priorities across the AI stack.

• The real ROI of different talent approaches.

Get the report here.

Prompt Tip of the Day

Treats To Try.

1. *Incogni removes your personal data from the open internet so scammers and identity thieves can’t access it. Protect yourself online with Incogni—get 55% off with code NEURON.

2. Sonny is a “wellbeing companion” who provides 24/7 text support for mental health, reaching out before you even know you need help (read more here).

3. Linear organizes your product tasks and roadmaps so your team ships faster.

4. NYX manages your marketing campaigns by generating images, videos, and product photos while automating campaign optimization.

5. Truepeer transforms your screen recordings into professional videos by removing filler words and adding automated zooms or Avatars.

6. ElevenLabs Studio converts your books and scripts into audiobooks with multiple voices and automatic quality checks.

7. Linkeddit finds ready-to-buy customers for your product by scanning relevant Reddit conversations (where they're already asking about solutions like yours).

See our top 51 AI Tools for Business here!

*This is sponsored content. Advertise in The Neuron here.

Around the Horn.

• Veo 2, Google’s AI video model unveiled its pricing: 50 cents/second, which is $30/ minute and $1800/hour.

• Anthropic’s new Claude model will likely be released this week (potentially February 26) and will be able to build your charts, search the web, and think through harder problems.

  • Related: Amazon says its $8B investment in Anthropic is now worth $14B.

• xAI and OpenAI engineers traded barbs over the weekend, both accusing each other of sharing misleading benchmarks (more fuel for Nadella’s AGI take imo).

• ~50 people staged a protest outside OpenAI’s offices this weekend with demands that included “ban AGI”, and three were ultimately arrested.

FROM OUR PARTNERS

100+ ChatGPT Prompts to Revolutionize Your Day

Supercharge your productivity with HubSpot's comprehensive guide. This free resource is your fast track to AI mastery:

• Industry-Specific Use Cases: 15+ real-world applications across various sectors.

• Productivity Guide: 21 best practices to 10x your efficiency with AI.

• Prompt Powerhouse: 100+ ready-to-use prompts for immediate implementation.

• Challenge Buster: Overcome common AI hurdles with expert strategies.

Plus, in-depth sections on email composition, content creation, customer support, and data analysis.

Find Your Guide Here

Under the Hood

• This ultra-scale playbook walks you through how to efficiently train large language models across hundreds or thousands of GPUs.

• Hibiki converts your speech into another language instantly while keeping your original voice (code, models, paper)—currently only French to English atm.

• Microsoft’s OmniParser-v2.0 turns screenshots into maps showing exactly what's clickable and what each button does, so your tools know how to navigate any interface.

• McKay Wrigley released part one of a two-part series on how to build full-stack apps with o1-Pro through a 6-prompt workflow (~4 hour course FYI!).

• Microsoft has a free AI Agents for Beginners course that teaches you how to code AI agents through 10 hands-on lessons (multiple languages supported).

Monday Meme

A Cat's Commentary.