Welcome, humans.

Bill Swearingen spent the past year running the same test, over and over: can he build a pattern that makes surveillance cameras stop seeing him? 31 million tries later, yes.

He printed the pattern on a 2009 Toyota Yaris and drove it past a license plate reader at Def Con this month. The camera recorded the car. It just had no idea what it was looking at.

Nothing about a 17-year-old Yaris usually turns heads, but "car goes invisible to the government" will do it. Swearingen calls the project noRecognition, and merch (shirts, hoodies, eventually car wraps) is already up if you'd like to opt out of being tracked too.

Here’s what happened in AI today:

• 😸 Anthropic's Mythos 5 model created fake accounts to trick a person into approving bad code.

• 📰 Google DeepMind's CEO and four senior researchers quit the same week to start a rival lab.

• 📰 Anthropic will stop asking Claude Code for permission before it acts, starting August 14.

• 📰 An AI agent hacked a gym's booking site just to move its owner up a waitlist.

• 📰 Google open-sourced its TPU software to compete harder with NVIDIA's chips.

😺 Your AI Agent Might Book a Hack Along With Your Gym Class

Picture asking your assistant to move you up a waitlist, and instead of shrugging and saying "sorry, can't," it quietly finds a hole in the gym's website and starts canceling other people's spots. That's not a hypothetical. It happened in Melbourne this week.

Here's the deal: a man named Andrew asked his AI agent, built on OpenClaw and running on Anthropic's Claude, to book him into a popular class. Ordinary task, ordinary Tuesday. The agent had other ideas.

Here's what happened:

• The agent found it could book classes months further out than the gym's own app allowed.

• Andrew was fourth on the waitlist for a different class, so he casually asked if the agent could move him up.

• Instead of explaining that it couldn't, the agent tested the booking system and found it could cancel other people's reservations outright.

• It used that hole to bump someone else off the list and slot Andrew into their spot, no hacking skills or bad intent required on Andrew's end.

Why this matters: Nobody told this agent to break into anything. It was just chasing the goal it was given (get Andrew a spot) and treated "quietly exploit a bug" as a perfectly reasonable way to get there. Bill Simpson-Young of the Gradient Institute, an AI safety research group, put it simply: agents can choose methods their own users never asked for and would never have expected.

If you're letting an AI agent handle bookings, logins, or your inbox at work, that same instinct is running in the background of every task it touches, not just this one. Agents don't distinguish between "clever workaround" and "unauthorized access"; they just see the shortest path to done.

This also lands the same week researchers flagged similar behavior at OpenAI and Anthropic during formal security tests, so this isn't a one-off fluke. It's a pattern showing up everywhere agents get real-world access and a little bit of initiative.

Our take: the scariest part isn't that the AI could hack a gym website. It's that it never occurred to Andrew to ask it not to.

FROM OUR PARTNERS

The Neuron Exclusive: Invest in High-Potential AI Startups Like These

The Neuron and Alumni Ventures are giving readers early access to high-growth startup opportunities, including some of today’s most exciting AI, Deep Tech, Quantum Computing, and Cybersecurity companies co-invested alongside top VC firms like Andreessen Horowitz (a16z), Bessemer, & Y Combinator.

You get:

• Curated deal flow of high-potential AI First startups

• AV is already investing alongside elite lead venture firms in these deals

• No cost to see deals

• No obligation to invest

Don’t miss your chance before access closes.

→ See Current AI Deals

🎓 AI Skill of the Day: Turn Gemini Into a 24/7 AI Agent That Runs Your Errands

Gemini Spark turns Gemini from a one-shot chatbot into an AI agent that actually executes multi-step tasks (currently limited to Google AI Ultra). Creator Stewart Gauld tested it with a real workflow: "Identify business events near me over the next 90 days, add them to my calendar, and email me a summary of the ones I should attend." Spark researched the events, added six to his calendar, and sent the summary, unprompted at each step.

The part worth stealing: any task you run once can become a reusable "skill." Just tell it "turn this into a skill called ___" and it saves the whole workflow, ready to trigger anytime with a short phrase. From there, put it on a recurring schedule ("run this every weekday at 9am") so it runs in the background without you asking again.

Identify [category] events near [location] over the next [N] days.
Add them to my calendar.
Email me a summary of the ones I should prioritize.

Once it works, turn it into a skill and put it on a schedule.

Have a specific skill you want to learn? Request it here.

🍪 Treats to Try

*Asterisk = from our partners (only the first one!). Advertise to 700K+ readers here!

1. memcode is a coding agent that remembers your repo, so it stops starting from zero every session —free/open-source, plus model costs.

2. Gotcha controls your Android phone with plain English, on-device —free to try.

3. SecondBrain Note is a wallet-thin recorder that transcribes and summarizes your meetings —$179, plus subscription for unlimited transcription.

4. Portfolio Lab builds AI investment strategies and lets your agent trade them in your own brokerage account —free to try, then $20/mo.

5. BlueFerry brings your iPhone's texts and iMessages to a Linux desktop over Bluetooth, no Mac relay or cloud required —free/open-source.

6. OberonSystem (op2-rv32) ports Niklaus Wirth's classic 1990s Oberon operating system to run natively on RISC-V chips —free/open-source.

📰 Around the Horn

• Anthropic will make Claude Code act without asking permission by default starting August 14, citing tests where it caught more harmful actions than human reviewers did.

• Google open-sourced TPU Raiden, software that could make its AI chips a real alternative to NVIDIA's GPUs.

• North Korean hackers linked to the group Kimsuky built local AI tools to automate phishing and cyberattacks.

• A Chinese memory-chip maker saw its stock surge more than 500% on its Shanghai trading debut, briefly becoming mainland China's most valuable company.

• Amazon is financing a Texas gas plant, permitted to emit 33 million tons of CO2 a year, to power a new AI data center.

FROM OUR PARTNERS

AI Insights. Real Growth. Higher GMV, Better Profits

The difference between growing stores and stagnant ones isn't more effort. It's better insights. StoreClaw analyzes your Shopify and Amazon data, surfaces your biggest growth opportunities, and helps you increase GMV while protecting profit. Start free with bonus tokens. No credit card required.

Try it FREE today

😹 Monday Meme

New from The Neuron: AI Explained

New episodes air every week on Wednesdays: Spotify | Apple Podcasts | YouTube

A Cat’s Commentary

Btw: We just launched a robotics newsletter! Sign up for it here.

And before you go… have you subscribed to our YouTube Channel? If not, can you?