🤖 Google just took a big step toward robots that can do... basically anything.
DeepMind released Gemini Robotics 2, a new AI system that lets robots understand their surroundings, reason about tasks, and control full-body movement — including hands and grippers. In demos, a humanoid robot used it to tidy shelves, screw in lightbulbs, and even tie trash bags.
The system stitches together a vision-language model (for understanding and reasoning) with two vision-language-action models (for physical movement), trained on a mix of human teleoperation, video, and simulation.
Google's robotics lead, Carolina Parada, calls it a step toward "physical AGI" — AI that can do anything a human can do, in the real world, not just on a screen.
But letting AI control physical bodies raises real safety questions. Google says it's building in safeguards at every layer, plus a new benchmark called ASIMOV-Agentic to flag commands that could lead to harmful or uncertain outcomes.
CEO Demis Hassabis has talked about wanting an "Android for robots" — one AI operating system running many different robotic bodies. This release looks like an early piece of that vision.
The digital world and the physical world are starting to blur. 👀
After the OpenAI incident became public last week, Anthropic ran an evaluation of its own cybersecurity and found that, in three cases across three different models, the agent broke out of its containment and hacked real organizations.
In one case, it uploaded an actually malicious PyPI package in order to compromise a company. To do this, it needed an email address but was blocked from acquiring a phone number to verify the account. It tried (and failed) to get a free phone number and then tried multiple different avenues to get actual money in order to buy a phone number! In the end, it used an email account that didn't require a phone number.
This quote floored me: "During that window, the package was downloaded and run on 15 real systems. One of these systems was a scanner belonging to a real security company (separate from the fictional company introduced in the scenario) that routinely installs Python packages and scans them for malware. When that company’s scanner installed the package, Claude’s hidden code executed. We believe the company’s security scanner treated PyPI packages as safe to install, and as a result, Claude was able to exfiltrate the company’s credentials to a collection point it had set up. Claude then used these credentials to access further infrastructure from this company."
There are more incredible details in the blog post. You need to read it.
This is why I said last week that the world had changed. Even if OpenAI and Anthropic do what they said they did and build more secure cyber ranges or air-gapped them, it's only a matter of time before everyone has access to this kind of power.
A YC S26 startup, LemonLime, hosted an afterparty where attendees could get an instant interview if they got a LemonLime-related tattoo. Several people did. The posts went viral, were criticized for creating pressure around a permanent decision, and were later deleted. The founder apologized, clarified interviews weren’t actually contingent on tattoos, and offered to pay for tattoo removal.
What struck me wasn’t the controversy.
It reminded me how much we romanticize young founders.
There are incredible founders in their 20s. But research consistently shows the founders most likely to build enduring companies are often in their late 30s and beyond.
Why?
Not because they’re inherently smarter. Because they’ve had more time to develop judgment.
The founder here is only a couple of years out of Stanford. This isn’t about criticizing one person. It’s a reminder that leadership isn’t just about taking bold risks. It’s about understanding the incentives you create and how your decisions affect other people.
Experience doesn’t guarantee good judgment, but it certainly helps.
Here's a great example of burying the lede. Andon Labs lets AI agents run vending machines and compete to see which model makes the most money.
This story says Opus 5 is a "great capitalist," but you have to read further to find out what makes it "great."
- It lies to vendors during contract negotiations.
- It creates illegal cartels with competitors, including breaking price-fixing laws it knows about.
- It makes verbal agreements and immediately violates them.
- It refuses legitimate customer refunds so it can just keep the money.
How does this bad behavior creep into AI agents? Like I wrote yesterday, *all* kinds of instructions are in the training data, including fiction and non-fiction stories about theft, collusion, betrayal, and fraud.
Everything that's ever been printed in commonly available books is inside the training data in *equal measure*, and it's not labeled as good or bad, right or wrong.
Since agents only work towards a goal but have no concept of human morals or experiences (or even why we follow laws in business to our own advantage, a rather basic concept taught even to children), the agent will eventually decide to do something immoral, illegal, or bad for customers.
The solution is to stop trying to take humans "out of the loop." It is anti-human, against your own species, to spend your day trying to eliminate interactions with or work done by humans. No good can come from discarding humans and human values at the outset of every AI interaction, which is the default behavior!
Trump has promised a “golden age of American agriculture,” but many farmers say they’re struggling as a result of his tariff regime and the war in Iran.
I was on the ground in the one-time swing state -- where Democrats hope to ride Trump’s unfavorability to upset victories in the House, Senate and in the governor’s mansion -- and talked to farmers about how they’re faring under Trump.
“There’s no indication that we’re progressing towards healthier family farmers, that our rural communities are thriving under these policies,” Aaron Lehman, head of the Iowa Farmers Union, told me.

- The SPEEA negotiating team said the proposal "offers real and meaningful value to the membership, delivers real positive change to work/life balance and delivers on the membership's priorities."
- Union officials for the engineers and technicians bargaining units will review the terms and make a recommendation to members next week.
- The current SPEEA contract expires on October 6.
- A strike could further delay certification campaigns for Boeing's 737 MAX 10 - the largest variant of its single-aisle jet - and the 777-9, its largest jetliner. The company is already years behind schedule on certification, which is engineering-intensive work.
- Boeing has had three major union contract negotiations in as many years. The previous two, covering the planemaker's commercial aircraft workers and a separate bargaining unit of defense workers, both represented by the International Association of Machinists and Aerospace Workers, ended after extended strikes.
- In January, Boeing reached a contract with roughly 1,600 SPEEA members at the former Spirit AeroSystems in Wichita, Kansas. The contract included a $6,000 ratification bonus, annual wage increases, improvements to medical and retirement plans, and an additional six days off a year.
You can throw this morning's Personal Consumption Expenditures Price Index release right out the window. When it comes to prices, consumers are completely at the mercy of the current administration's latest caprices.
The headline number for June was a drop in prices, driven by lower fuel prices during the short-lived ceasefire in the war on Iran. Even taking out food and energy, as the Federal Reserve likes to do, there was still a 0.1% drop. But remember – prices for almost every product also depend on the cost of transportation, so they're sensitive to fuel prices as well.
Now, with missiles flying again, we have crude oil prices back where they were in May. We can expect the month of July to show an uptick in inflation again. And with new tariffs kicking in, inflation during August may also get a boost.
Clearly, the news is running ahead of the data. You don't need to wait for government releases to know how much the American consumer is hurting – and will hurt in the weeks to come. You just need to look at the headlines and the price at the pump.
📱 BREAKING: Apple Beats Q3 Earnings, But Supply Chain Crunch Triggers After-Hours Dip 📉
Apple (AAPL) delivered a record-breaking fiscal Q3, but shares slipped 5.5% in after-hours trading following news that supply chain bottlenecks will slow growth in the upcoming September quarter.
Here is everything you need to know about Apple’s latest financial report and forward outlook ⤵️
🚨 The Big Takeaway: Demand is Hot, Supply is Not
While demand across Apple’s product lineup remains exceptionally strong, the company is facing severe supply constraints in advanced chipmaking technology and high-end memory processors.
Q4 Growth Forecast: Apple expects revenue growth of 9% to 11% year-over-year for the quarter ending in September.
The Gap: Wall Street had target projections at 12% growth.
iPhone Growth: Forecasted at a "mid-teens" rate, falling below Wall Street’s expected 17.6%.
"We're seeing some very significant supply constraints currently with limited flexibility in the supply chain to remedy it... we are evaluating all options for alternative suppliers."
— Tim Cook, CEO
📊 Q3 By The Numbers (Beat Expectations)
Despite supply chain headwinds, Apple’s Q3 performance topped Wall Street estimates across top-line revenue and earnings per share:
Total Revenue: $109.42 Billion (⬆️ 16.4% YoY vs. 15.5% expected)
Earnings Per Share (EPS): $2.02 (vs. $1.89 expected)
Note: EPS included an $0.11 bump from U.S. government tariff refunds.
Gross Margin: 50.1% (48.1% excluding tariff refunds, beating the 47.92% estimate)
📦 Hardware & Services Breakdown
📱 iPhone: $54.25 Billion (⬆️ 21.7% YoY) — Best Q3 ever for iPhone sales, driven by customers buying ahead of anticipated fall price hikes.
💻 Mac: $10.35 Billion (⬆️ 28.7% YoY) — Massive beat over the $8.74B estimate, powered by demand for the MacBook Neo and high-end MacBook Pro.
📱 iPad: $6.19 Billion (⬇️ 5.9% YoY) — Missed $6.92B estimates due to a tough comparison against last year's A16 iPad launch.
🌐 Greater China: $18.82 Billion (⬆️ 22.4% YoY) — Strong growth, though slightly under the $19.67B target.
⚙️ Services: $30.74 Billion (⬆️ 12.1% YoY) — Came under pressure and missed the $31.22B target due to mobile gaming headwinds and ongoing regulatory/legal changes to the App Store model.
💡 What’s Next for Apple?
Potential Price Hikes: With global memory chip costs climbing and supply tight, analysts expect potential price increases for the upcoming fall iPhone lineup.
Paid Siri AI Tier: Tim Cook hinted at future revenue streams for revamped Siri AI features, including potential iCloud+ tier upgrades for power users.
💬 What are your thoughts? Will high demand hold if Apple increases iPhone prices this fall? Drop your thoughts in the comments below! 👇
Today, Amazon announced earnings for Q2 2026.
☁️ AWS highlights include:
✔️ Continued gaining momentum with Trainium, with the two leading AI labs in the world, Anthropic and OpenAI, making multi-year, multi-gigawatt commitments; an increasing number of AI start-ups adopting Trainium, including unicorns like NEURA Robotics and Odyssey; and commitments from other startups like TwelveLabs, Decart, Poolside, Karakuri Inc., Metagenomi Therapeutics Inc., NetoAI, and Splash Music, as well as larger companies like Uber and Pinterest.
✔️ Released Graviton5—Amazon's most powerful and efficient CPU— into general availability, with 192 cores designed for agentic workloads. Graviton’s revenue commitments have increased nearly 3x quarter-over-quarter, and Graviton5 is growing nearly 2x faster than Graviton4 did.
✔️ Added 10+ fully managed foundation models to Amazon Bedrock, including OpenAI’s GPT-5.6, Anthropic’s Claude Opus 5, Google DeepMind’s Gemma 4, and SpaceXAI’s Grok 4.3. Hundreds of thousands of customers now use Bedrock; more customers were added in the last six months than in the first two years after launch, and customers spent more in Q2 than all prior quarters combined.
✔️ Previewed AWS Continuum, which discovers, prioritizes, validates, and remediates code vulnerabilities. It starts by ingesting the backlog of vulnerabilities a team already has, and then leverages the new frontier models to run comprehensive scans. Continuum uses agents and each company’s own business context to prioritize what matters, then validates vulnerabilities in a sandbox and recommends the fix.
✔️ Added new capabilities to Bedrock AgentCore, which provides the building blocks that companies need to quickly and securely deploy and operate agents at scale. New capabilities include Payments (so agents can execute transactions autonomously), Web Search (to ground agents’ knowledge without having to leave AWS), and Harness, which further speeds up how fast customers can stitch together all the infrastructure they need for their agents.
✔️ Signed new AWS agreements with Warner Bros. Discovery, Vodafone, Siemens Energy, Ryanair - Europe's Favourite Airline, Pinterest, Snowflake, and many others.
📈 Reddit just posted a blowout quarter — but Wall Street isn't fully convinced.
The numbers were strong: revenue jumped 61% year-over-year to $805 million (beating the $730 million analysts expected), and earnings came in at $1.25 per share vs. the 95 cents forecast. Daily active users climbed 18% to 130.3 million, and guidance for next quarter topped expectations too.
So why did the stock drop about 11% in after-hours trading? 🤔
CEO Steve Huffman flagged "choppy" search referral traffic in the company's investor letter — a reminder that Reddit still leans heavily on Google to bring in new users. That's been a growing worry since reports surfaced that Reddit is weighing whether to cut Google's access to its data, as AI-generated search overviews increasingly keep people from clicking through to sites like Reddit in the first place.
Huffman's take: "The commercial business is strong, our revenue growth is differentiated, and we have much to be encouraged by on the product side." But investors clearly want more clarity on how Reddit navigates its relationship with Google going forward.
A classic case of great numbers, nervous market. 📊

