An earthquake hit the land of robots last week.

Google released its Gemini Robotics 1.5 AI model, and this one is a lot more scalable than any of its predecessors.

Until now, in AI robotics, any minor variation in a robot’s physical form would require a completely new, fine-tuned brain. But Google’s latest release solved this problem by making a more general-purpose “brain” that can be used in robots of any shape.

Basically, it can do a wide array of jobs, from washing dishes to working at a nuclear power plant.

On top of that, it can figure out how to adapt when things go wrong. Imagine a robot with a broken leg learning to walk with the other three legs, or a robot hand with one of its motors jammed, figuring out how to still grab an apple.

This fundamental shift opens up the floodgates of consumer robotics and might usher in an era where living with robots is as common as logging onto the internet.

Which is pretty cool. Until you think about the “Terminator” movies, or “Robocop.”

In today’s edition, we’re digging into some of these breakthroughs, along with consequences for the job market, and how policymakers are playing catch-up.

That exchange was from the launch video from Google last week, showing off its impressive “agentic” capabilities.

Three years ago, when ChatGPT was first released, it would answer quite a lot of questions with answers like “As an AI language model, I do not have the ability to browse the internet or access real-time information.” We’ve obviously come a long way from that.

Now, robots are going to evolve at similar speeds, going from simple sorting of objects to actually understanding the world we live in.

Another instance of robots “learning” is from Skild AI. A new demo video shows their robot dog adapting to broken and modified limbs in real-time.

Right now, the majority of these robots’ real-world applications have been in fulfillment (like at Amazon centers) and manufacturing.

And that’s where the money has gone. AI-powered warehouse and industrial automation got over 70% of robotics startup funding in Q1 2025, accelerating the rollout of collaborative robots (“cobots”) for tasks like pick-and-place, palletizing, sorting, and machining with just-in-time learning capabilities.

Amazon now operates more than 1 million robots across its operations network, making it the largest user of robotics in the world. These robots are used in nearly 300 fulfillment centers, working alongside humans to perform repetitive tasks.

Amazon also has about 1.5 million people working in its facilities.

As these robots are being rolled out, we can see the direct impact on Amazon’s labor force.

Last year, Amazon averaged about 670 human employees per facility, which was the lowest in the past 16 years, the Wall Street Journal reported.

But as robots evolve, they’re breaking out of warehousing and manufacturing.

Already, robots are being used more and more for “security” (like in “I, Robot”).

And that raises huge policy questions about what kind of authority we want to hand over to our robot overlords.

In last week’s edition, we saw firsthand the dangers of chatbots and allowing intelligent robots to occupy our world.

As these “ready-made brains” go from being fun demos to real-world use, we need to pay attention to which sectors it is being rolled out in and how lawmakers react.

State lawmakers are starting to take action.

But, as usual, the law moves like an iceberg trying to catch a rocket ship.

Still, there are a handful of bills that deserve some attention:

The themes that worry lawmakers are clear: liability, workforce disruption and weaponization.

But without a federal framework, the rulebook for dealing with smart robots remains thin.

The eyes of the nation turn to California: After a prolonged dispute with AI companies, California Gov. Gavin Newsom signed SB53, the first law in the country to force AI companies to show their safety protocols, Politico reports. The law also paves the way for the public to report major safety issues to state officials, complete with whistleblower protections for employees of AI companies.

The Ballad of Tilly Norwood: The actors union in Hollywood is up in arms over an AI “actress” named Tilly Norwood, NPR reports. So far, the avatar of a brunette actress can only deliver monologues facing a camera in social media posts. But SAG-AFTRA President Sean Astin (known for his roles as Rudy and Samwise Gamgee) says actors are concerned the avatar is the first step toward replacing humans on screen, which would violate an agreement between the union and movie studios.

Maybe not the best kind of sandbox: Arizona is one of six states where the Trump administration will test an AI-powered program designed to help the federal government deny care to Medicare patients, KFF Health News reports. Starting in January, the pilot program will weed out low-value services through a process known as prior authorization. It’s not a popular practice. Earlier this year, Mehmet Oz, a physician and former TV personality who now serves as administrator of the Centers for Medicare and Medicaid Services, said prior authorization “erodes trust” when it’s used by private health insurers. Also this week, the White House announced a plan to spend $50 million on using AI to gather better data and guide pediatric cancer research.

But can it get there in 30 minutes or less?: The food delivery service DoorDash is testing out an autonomous robot known as “Dot” in Tempe and Mesa, per the Associated Press. The robot is about the size of a big baby stroller and can go as fast as 20 mph on streets and sidewalks.

DIY agents: Arizona State University will have a new class next spring taught by Will.i.am of the Black Eyed Peas. He is going to teach a course on AI, “The Agentic Self,” where students will create AI universal learning companions to aid in their studies and professional careers. The rapper said the course “represents a solution to AI replacing human jobs.” He made the announcement alongside ASU President Michael Crow at a United Nations conference in New York, ASU News reports.

Quantum at the top: The White House has placed quantum science and artificial intelligence at the summit of federal R&D priorities for FY 2027, directing agencies to channel funding toward breakthroughs that tie directly to health, energy, defense and space. The memo frames AI as both a field and a force multiplier, accelerating discoveries in quantum and beyond, while quantum gets a dual push — fundamental science and engineering maturity.

Accenture cuts 11K Jobs: Over the past three months, Accenture has slashed more than 11,000 roles — cutting its global headcount from 791,000 to 779,000 — saying many workers “cannot be retrained” fast enough for AI-era needs. The company booked $615 million in severance and restructuring costs this quarter, with another $250 million expected next quarter under an $865 million transformation plan.

OpenAI goes proactive: OpenAI is rolling out a new “proactive insights” feature that jumps in before you even start chatting. Instead of just answering questions, ChatGPT will now suggest ideas, context, or next steps — a shift from reactive assistant to always-on guide.

Five new Stargate data centers: OpenAI—alongside Oracle and SoftBank—announced five new U.S. sites under the Stargate infrastructure initiative, pushing the project toward its $500 billion / 10-gigawatt goal ahead of schedule. The expansion brings Stargate’s planned capacity to nearly 7 gigawatts, and is expected to create over 25,000 on-site jobs, plus many more across the U.S.