💥 AGI Debate Breaks Into the Open

A clash between AI heavyweights Yann LeCun and Demis Hassabis has reignited a big question: is general intelligence even real? LeCun argues that what looks like human flexibility is actually shaped by our biology, senses, and the need to be efficient, adding fuel to an already lively debate. Let’s dive in!

In today’s insights:

LATEST DEVELOPMENTS

Evolving AI: A public clash between two AI pioneers reignited the debate on whether general intelligence is real.

A social media exchange between Yann LeCun and Demis Hassabis sparked renewed debate about general intelligence. LeCun argues human intelligence looks flexible but is tightly constrained by biology, sensors, and efficiency limits. Hassabis counters that generality refers to theoretical capacity, not performance, and says human brains and modern AI models approximate general computing systems, even if resources are bounded.

This back and forth shows a real split in how AI labs decide what to build next. Teams that side with LeCun tend to focus on systems that excel at specific jobs like vision, language, or planning, then ship products fast. Teams closer to Hassabis keep pouring money into large models that learn many skills at once, betting on reuse across research, science, and consumer tools. That choice shapes hiring, compute spend, and timelines for things users actually touch, from coding assistants to drug discovery platforms. The fight over AGI is really a fight over where the next decade of AI effort goes.

Smart Brevity is the methodology behind Axios — designed to make every message memorable, clear, and impossible to ignore. Our free toolkit includes the checklist, workbooks, and frameworks to start using it today.

Access the free resources

Evolving AI: Alphabet is buying Intersect to secure data center capacity and energy supply for AI growth.

Alphabet plans to acquire Intersect, a data center and energy infrastructure firm, in a $4.75B cash deal plus assumed debt. Intersect will stay operationally independent while working closely with Google’s technical infrastructure teams. The acquisition aims to speed up co-located data center and power generation projects, including sites in Texas. Certain California and Texas assets are excluded and will remain under Intersect’s current investors. The deal is expected to close in the first half of 2026.

AI is starting to look less like a model race and more like an electricity-and-land race: Alphabet is paying $4.75B for Intersect so Google can line up new data center capacity with new on-site generation, instead of waiting in the grid queue like everyone else. Zooming out, it’s another sign hyperscalers don’t just want clean energy contracts, they want direct control over the projects feeding their AI load, and that reshapes who wins permits, where new campuses get built, and how fast the next wave of AI features can actually ship.

ChatGPT is getting its own Spotify Wrapped moment. OpenAI is rolling out “Your Year with ChatGPT,” a personalized, end-of-year recap that highlights how you’ve used the chatbot over the past year. The experience leans into playful visuals and custom “awards” that reflect your habits. According to OpenAI, it’s designed to be lightweight, privacy-forward, and fully user-controlled, giving people a fun snapshot of their year with AI without overstepping on data use.

If you update your app, you should be able to see the function. You can also trigger the experience and ask ChatGPT for “Your Year with ChatGPT”.

Evolving AI: OpenAI says AI browsers may never fully escape prompt injection attacks.

OpenAI says prompt injection attacks, where hidden instructions manipulate AI agents, are unlikely to disappear. Since Atlas launched in October, researchers showed how simple text in documents or emails could alter agent behavior. OpenAI is responding with layered defenses, faster patch cycles, and an automated attacker that tests new exploits in simulation before real misuse appears.

Agentic browsers are basically getting “hands” on your behalf, and prompt injection is the reminder that the web is full of content that can talk back. If a hidden line in an email, doc, or webpage can steer an agent into sending messages or touching payments, the safe default shifts from “let it roam” to “give it a narrow job, minimal logins, and make it ask before it acts.” OpenAI’s own stance is that this is more like fighting scams than patching a bug, so the winning workflow looks like least-privilege access plus fast defense updates, not blind trust.

It just surpassed a $1B valuation, joining private US companies like SpaceX and OpenAI. Unlike those companies, you can invest in EnergyX today. Industry giants like General Motors, Eni, and POSCO already have. Why? EnergyX’s patented tech can recover up to 3X more lithium than traditional methods. Now, they’re readying 100,000+ acres of lithium-rich Chilean land for commercial production. Join 40k+ people as an EnergyX investor.

Invest in EnergyX Today

_{Energy Exploration Technologies, Inc. (“EnergyX”) has engaged Nice News to publish this communication in connection with EnergyX’s ongoing Regulation A offering. Nice News has been paid in cash and may receive additional compensation. Nice News and/or its affiliates do not currently hold securities of EnergyX.}

_{This compensation and any current or future ownership interest could create a conflict of interest. Please consider this disclosure alongside EnergyX’s offering materials. EnergyX’s Regulation A offering has been qualified by the SEC. Offers and sales may be made only by means of the qualified offering circular. Before investing, carefully review the offering circular, including the risk factors. The offering circular is available at invest.energyx.com/.}

QUICK HITS

👔 Work is changing everywhere as AI moves from experiment to expectation.

🎓 AI has become the norm for students. Teachers are playing catch-up.

🆚 New York Times reporter files lawsuit against AI companies.

🤥 AI app apologises over false crime alerts across US.

*partner link