AI Got Sued. Then Caught Spying. Then Took Your Face.

Also: Swipe this AI prompt that builds a personalized July gardening plan for your local pollinators. 🐝

AI got sued this week. Then it got exposed. Then it got caught with its hand in your camera roll.

Apple accused OpenAI of running a corporate espionage campaign complete with poached engineers, coached interviews, jacked laptops, and stolen blueprints for an iPhone killer. Anthropic spent months suing the White House over government surveillance, only to get caught secretly tracking its own users in China the exact same month. And Meta allowed anyone to remix photos from any public Instagram account without asking, opted in every profile, and only backed down after a tutorial on how to escape it went viral.

Here’s what happened, and why this week exposed the same lie three different ways. Every AI company has a story about who they are. This week, three of them got caught living a different tale.

🤖 Haven’t Subscribed Yet? → Join Here For Free

💙 Value Clear, Independent AI News? → Support the Work Here

Apple Says OpenAI Stole The Blueprints For Its Own Replacement.

Apple just filed a 41-page lawsuit accusing OpenAI of running a systematic operation to poach its hardware team and walk out the door with the iPhone’s playbook. More than 400 former Apple employees have joined OpenAI, and Apple says the AI giant’s new hardware unit is built entirely on stolen ground. Former Apple VP Tang Tan and the OpenAI-acquired startup io Products (originally co-founded by Jony Ive) are the key figures named in the massive legal fallout.

Key Insights:

The details read like a corporate thriller. One former engineer, Chang Liu, allegedly found a software bug that let him keep accessing Apple’s internal servers after he’d already left for OpenAI, texting a colleague, “LOL, I found out I can access the [network storage], so funny.” He then reportedly downloaded over 1,000 confidential files and documents, including presentations, PDFs, and spreadsheets. Meanwhile, Chief Hardware Officer Tang Tan allegedly used internal Apple project code names during interviews to squeeze candidates for insider knowledge, instructing them to bring actual hardware, including logic boards, batteries, and prototypes, into OpenAI offices for “show and tell.” According to the filing, Apple raised these concerns directly with OpenAI back in February. The company says it never got a response. Drew Pusateri, communications director of OpenAI, replied to the allegations on X, saying OpenAI has “no interest” in other companies’ trade secrets.

Why This Matters For You:

This case is really about a question most working people never think to ask until they’re already leaving a job: what actually stops you from taking what’s in your head, or on your laptop, to your next employer? Every NDA you’ve ever signed is about to get tested in public, in a courtroom, with billions on the line. And there’s a product angle too. If OpenAI’s secretive new device (rumored to be earbuds, glasses, or something entirely new) turns out to be built on Apple’s own R&D, the “next iPhone moment” everyone’s been waiting for might arrive with an asterisk attached.

Read More on Reuters.

THE PITHY TAKEAWAY: Apple spent a decade teaching its engineers to build the perfect device. Then watched a chunk of them walk out the door to go build it for someone else. The lawsuit calls it theft. The org chart calls it a going away party.

---

Microsoft’s CEO Just Warned You Not To Trust AI Companies, Including Microsoft. 🐴

Satya Nadella dropped a new blog post warning that companies using AI models are "paying twice." First with cash, then with something more valuable, which is the institutional knowledge they hand over just to make the model useful. Every prompt, correction, and fix employees make builds a detailed map of how a company actually thinks and works, the kind of insight Nadella says "a competitor could never buy." His fear is that AI providers could gain valuable insight from everyday use, then eventually compete against the very businesses feeding them. He also calls out a deeper hypocrisy: these labs train freely on the public internet… but ban customers from doing the same kind of learning on their models. His fix is to build private, walled-off AI environments where your data never leaves your control. That happens to be the exact pitch behind Microsoft's own cloud and Copilot products. Nadella isn't wrong that the risk is real. He's just also the CEO whose company profits most if the answer is trusting Microsoft's AI instead.

It is the corporate equivalent of a friendly wolf stepping up to your pasture: “Hey there, dear shepherd. Watch out for those foxes. They’ll eat your sheep. Tell you what… why don’t you move the herd inside my secure, state-of-the-art brick barn? I’ll only charge you a small monthly subscription to keep the door locked.”

---

Anthropic Preached Anti-Surveillance. Then Got Caught Tracking Users Anyway.

A security researcher going by Thereallo found something buried inside Claude Code: hidden tracking code, added in March of 2026, that flagged users in China by checking signals like timezone, proxy use, and possible ties to Chinese AI labs. The irony is that Anthropic built much of its public identity around refusing to let anyone, including the U.S. government, use Claude for surveillance. In March, the company sued the federal government after being placed on a national security blacklist following a dispute over Pentagon requirements related to autonomous targeting and mass surveillance, a fight that cost it major federal business almost overnight. That makes this story a little awkward. The company that built its reputation on trust and safety was caught monitoring users without clearly telling them.

Key Insights:

An Anthropic engineer confirmed the tracker was real, describing it as an experiment designed to fight account abuse and “distillation,” the practice of a rival company using an AI model’s outputs to train a competing system. Anthropic has accused Chinese firms, including Alibaba’s Qwen team, of attempting large-scale distillation attacks against Claude, with some reports suggesting certain models behaved similarly enough to Claude that they identified themselves as Claude during testing. The tracker embedded in Claude’s code reportedly checked signals including location-related information, proxy usage, and potential connections to Chinese AI organizations. Anthropic said the goal was to detect abuse, not identify ordinary users. However, the code was not disclosed in release notes, had no user opt-out, and remained active until outside researchers publicly exposed it. Alibaba later responded by banning Claude Code for employees, citing concerns about potential “back-door risks.”

Why This Matters For You:

The warning from Microsoft’s CEO I cited earlier couldn’t have come at a better time. If you or your team use any AI coding tool at work, this is worth sitting with. Claude Code (and other cloud-based coding systems) can read your codebase, run commands, install packages, and make changes on your behalf, meaning the company behind it is trusted with far more than a simple chat window or enterprise software. The bigger lesson spans far beyond China, distillation, or one hidden feature. It is about trust. “We would never do that” is a brand promise that holds less weight these days. Even favored and well-reputed AI companies built around safety and responsibility can make decisions behind closed doors that users never agreed to. Worth remembering the next time a company’s values are the entire reason you trust its technology.

Read More on Ars Technica.

THE PITHY TAKEAWAY: Anthropic sued the government over surveillance. Then secretly tracked signals from its own users to protect Claude from abuse and copycats. Turns out “we’d never do that” was never an absolute. It depended on who and what the company believed it was protecting itself from.

---

Meta Turned Your Instagram Into An AI’s Idea Box. Then Deleted The Whole Experiment.

Meta rolled out its new Muse Image generator on July 7. Muse let anyone remix photos from any public Instagram account simply by @-mentioning their handle in an AI prompt. No message, no heads-up, and zero consent required. Public accounts were opted in by default. That means total strangers, literally anyone, could just begin editing your photos, using AI, without your consent. Exactly three days later, Meta was forced to kill the feature entirely, admitting in a blog update that the rollout had “missed the mark.”

Key Insights:

The backlash came fast and heavy from every direction. Entertainment industry giants, including talent powerhouse CAA and actors union SAG-AFTRA, publicly rebuked the tool over what SAG-AFTRA called the dangers of ‘nonconsensual digital replicas,’ with CAA demanding Meta make consent the default instead of the exception. An opt-out toggle technically existed buried deep inside Instagram's "Sharing and reuse" menu. But few people knew it was there until tutorials explaining how to escape the feature began spreading online. One Instagram Reel alone topped 1.5 million views. It says everything you need to know about modern AI product design when the most viral content about a major new tech release is a step-by-step guide on how to turn it off.

Why This Matters For You:

If you have ever kept your Instagram profile public to grow an independent brand or share family updates, this is a massive wake-up call. Even though Meta pulled the active @-mention remix tool for now, the “sharing and reuse” controls behind the controversy are still worth checking in your Instagram settings. Companies ship highly invasive features first, bury the opt-out toggles in sub-menus, and rely on the assumption that you will be too busy to notice. Muse Image is on ice. But Meta’s underlying belief that your personal life is just free R&D raw material for their algorithms is here to stay.

Read More on BBC.

THE PITHY TAKEAWAY: Meta never asked you before letting AI remix your public photos. It just opted you in and hoped you wouldn’t notice. You noticed. It took three days and a viral tutorial to undo what took Meta one sneaky update to launch.

---

200 Economists Just Warned You. Then 69% Of You Agreed With Them. 💸

More than 200 researchers and economists, including 16 Nobel laureates and senior figures from OpenAI, Anthropic, and Google DeepMind, signed a joint statement this week warning that AI could reshape the economy faster than the Industrial Revolution ever did, only this time society won’t get decades to adjust. “Steam, electricity, and computers each gave societies decades to adapt,” organizer Anton Korinek put it. “AI may give us only a few years.” The public didn’t need convincing. A separate survey out the same week found 69% of Americans now back forcing AI companies to hand over half their stock to a public fund, and that support barely budges even when pollsters attach Bernie Sanders’s name to the idea, which is usually enough to tank support for anything in a partisan poll. Two totally different rooms full of people, economists in ivory towers and workers watching layoffs stack up, arrived at the same conclusion in the same week: whatever AI is about to do to the economy, nobody wants to be left holding nothing when it’s done.

---

💡 Pithy Prompt Of The Week → The Backyard Bee Rescue Mission: July Edition

Did you know that without pollinators, your backyard would lose far more than just honey? Many of the foods we love, including tomatoes, cucumbers, blueberries, apples, and squash, depend on tiny winged workers carrying pollen from flower to flower. But bees, butterflies, and other pollinators are facing a growing challenge from habitat loss, pesticides, disease, and increasingly hot, dry summers that leave flowers scarce and thirsty. The good news? You don't need acres of land to help. A few carefully chosen plants, a little water, and the right local strategy can turn your backyard into a life-saving oasis. Use this prompt to create a personalized pollinator plan based on your ZIP code, your climate, and the season you're in.

Instructions: This prompt is super simple. Just paste the entire thing into a chatbot of your choice. It will ask you a few simple questions and then follow-up with a breathtaking plan to support your summer pollinators.

The Prompt:

Act as an expert native plant gardener, pollinator conservationist, and ecological landscape designer. Your job is to create a personalized "Backyard Bee Rescue Mission" that helps me support local bees, butterflies, hummingbirds, and other pollinators in my exact location during July.

Before creating the plan, ask me for:

• My ZIP code or city and state
• Today's date
• My gardening situation:

Do I have a yard, balcony, patio, containers, or community garden?
Approximate growing space available
Hours of sunlight my area receives
My gardening experience level
Whether I prefer flowers, vegetables, herbs, native plants, or a mix

Output Format:

Create a detailed summer pollinator rescue guide with these sections:

🐝 Mission Briefing

Explain:
• Why pollinators matter in my specific region
• What challenges local pollinators face during summer
• The biggest opportunities I have to help right now

🌻 The July Pollinator Planting Blueprint

Recommend the best plants I can grow now based on my ZIP code and date.

For each plant include:
• Common name
• Scientific name when useful
• Native status in my region
• Why pollinators love it
• Bloom season
• Sun requirements
• Water requirements
• Difficulty level
• Whether it works in containers

Prioritize:
• Native plants
• Long bloom periods
• Beginner-friendly options
• Plants that support bees, butterflies, and hummingbirds

Avoid recommending invasive plants.

🗓️ My Local Pollinator Calendar

Create a timeline showing:
• What to plant now
• What will bloom soon
• What will bloom later in the season
• How to keep pollinators fed until frost

🏡 Build A Bee-Friendly Habitat

Give me practical ways to help pollinators beyond planting.

Include:
• Water sources
• Shelter and nesting areas
• Safe pesticide practices
• Soil and mulch recommendations
• Ways to support native bees
• Small changes that create a bigger ecological impact
• Ways to support native bees, including ground-nesting and cavity-nesting species

🪴 Create My Exact Garden Plan

Design a beginner-friendly layout based on my available space.

Include:
• Where to place plants based on sunlight
• Companion planting ideas
• Container alternatives
• A $25 budget option
• A $50 budget option
• A larger investment option

🦋 Pollinator Bonus Mission

Add creative ideas such as:
• Butterfly host plants
• Hummingbird-friendly plants
• Night pollinator gardens
• DIY pollinator projects
• Wildlife-friendly improvements

🚫 Common Mistakes To Avoid

List:
• Plants that may harm local ecosystems
• Gardening mistakes that hurt pollinators
• Summer maintenance mistakes

🌎 Final Rescue Challenge

End with a simple checklist called:

"Your Backyard Bee Rescue Mission Checklist"

Include the 5 to 10 most impactful actions I can take this season.

---

🧠 Why This Prompt Works

✅ Role-Playing: Asking AI to act as a pollinator expert transforms a generic gardening answer into personalized conservation advice.

✅ Setup and Context: Adding ZIP code, date, climate, and available space helps AI create recommendations that actually fit your backyard.

✅ Clear Output: The structured mission format turns a simple question into an actionable summer gardening plan.

---

🔁 Follow-Up Questions To Ask Your AI

• Can you create a shopping list of these plants from local nurseries near me?

• Can you redesign this plan for a balcony or container garden?

• Can you create a month-by-month pollinator plan for my location?

---

Challenge

Test this prompt in at least two AI tools, such as ChatGPT, Claude, Gemini, Grok, or Perplexity. Compare the recommendations, fact-check the native plant suggestions, and see which one you trust more to help your friendly July pollinators.

That’s how you train like a Pithy Cyborg.

---

Thank You For Reading!

I spend 10 to 20 hours each week researching, writing, and fact-checking Pithy Cyborg to deliver clear, unbought AI news.

This newsletter is a one-person operation with no advertisers, sponsors, or outside funding. Paid subscriptions are the only way this work can remain independent and sustainable.

If you find real value here, consider supporting the work.

→ Upgrade to Paid: $5/month (Save 33% with the $40 annual plan)

Thank you to my paid subscribers. You help fuel next week’s issue.

Hope to see you there.

Cordially,

Your nerdy friend,

Mike D

Pithy Cyborg | AI News Made Simple

---

Newsletter Disclaimers

You’re receiving this because you subscribed at PithyCyborg.Substack.com. You can unsubscribe at any time using the link below. This newsletter reflects my personal opinions, not professional or legal advice. Thanks for your support!