AI Invented A War This Week. (The Other Two Stories Are Actually Worse.)

Also, an elite AI prompt to plan the perfect salad garden for crisp, homegrown greens.

AI invented a fake war this week. Then it broke your mind. Then it came for your paycheck.

Hundreds of AI-generated explosions, missile strikes, and fictional casualties flooded social media during this new Iran war, reaching millions of people who had no idea they were watching fabricated battle scenes. A new study in Lancet Psychiatry found that AI chatbots are quietly accelerating delusional thinking in vulnerable users, and a wrongful death lawsuit against Google is putting a human face on what that actually looks like. And the ServiceNow CEO warned that college grad unemployment could hit the mid-30% range within two years, right as Meta, Oracle, and Atlassian announced tens of thousands of layoffs to fund their AI ambitions.

Here’s what happened, and why this week felt different. AI has been theoretical for a long time. A productivity tool. A party trick. A promise. This week, it was a weapon, a manipulator, and a pink slip. Sometimes all three before breakfast.

💙 Enjoy This? → Support Independent AI Journalism!

Fake AI Bombshells Are Triggering Real-World Chaos.

As the war with Iran brings chaos to the Middle East, war is being fought on two fronts. One of them is entirely fictional. The New York Times documented over 110 AI-generated fake images and videos circulating across X, TikTok, Facebook, and private messaging apps during the opening weeks of the conflict. Fabricated missile strikes on Tel Aviv. A fictional USS Abraham Lincoln engulfed in flames. AI-generated “short films” of destroyed schools, invented child victims, and downed jets. Millions of people watched these videos. Most had no idea they were fiction. How many millions are still living inside a manufactured reality?

Key Insights:

These AI-generated wartime deepfakes surpass everything previous generations of propagandists could achieve. Tools like OpenAI’s Sora let almost anyone produce cinematic, Hollywood-quality war footage for nearly zero cost. Historical war footage is grainy, distant, and undramatic. These new fakes are visually spectacular, emotionally precise, and built to spread. The Times found the fabricated content overwhelmingly pushed pro-Iranian narratives, exaggerating military power and inventing victories. Analysts are now calling AI disinformation “a natural front“ in modern geopolitical conflict. X announced it would demonetize accounts posting unlabeled AI war content for 90 days, though enforcement has remained spotty. The fakes spread far enough and fast enough that President Trump addressed them directly this week, calling images of the USS Abraham Lincoln on fire “fake news“ generated by AI, and confirming the carrier was never touched. Trump went further, threatening media outlets and journalists whom he accused of reporting the false claims with treason charges and arrest. AI disinformation doesn’t just confuse the public. It destabilizes the people in charge of responding to it.

Why This Matters For You:

You don’t have to be a soldier or a policymaker to be a target of misinformation. Every person scrolling a feed during a geopolitical crisis is now inside the blast radius of AI propaganda. These fakes are engineered to appear more genuine than reality itself. Right now, they are winning. The honest question to ask yourself is not whether you have seen AI-generated war content. It is whether you would have known if you had.

Read More on The New York Times.

THE PITHY TAKEAWAY: The visceral war footage you’re watching on Facebook might not be real. The explosion you just shared might be fiction. The side you’re rooting for might have been chosen for you, by an algorithm, before you ever opened the app.

---

30% Unemployment. For New College Grads. Within Two Years.

ServiceNow CEO Bill McDermott dropped a bombshell this week, warning that unemployment among recent college graduates could easily reach the mid-30% range within two years, driven almost entirely by AI adoption. As the leader of the company that powers AI agents for thousands of enterprises and is itself deploying autonomous AI specialists at massive scale, McDermott isn’t speculating. And his bleak outlook doesn’t seem far-fetched when you see the Fed’s unemployment data among recent graduates. As of this morning, the Federal Reserve Bank of New York clocks recent grad unemployment at roughly 5.7%. Far worse, they gauge recent graduates’ underemployment at a staggering 42.5%. And the automation storm hasn’t fully arrived yet.

Key Insights:

The layoffs are already cascading. Meta just announced plans to cut 20% of its workforce, roughly 15,000 jobs. Oracle is preparing to slash between 20,000 and 30,000 positions to cover the financial strain of its AI data center bets. Atlassian cut 1,600 employees last week, explicitly to “self-fund AI investments.” Block is cutting nearly half its workforce. The pattern is identical everywhere. Companies grow revenue while shrinking headcount. AI agents quietly replace entry-level and white-collar roles that once defined a middle-class career path. ServiceNow itself now resolves over 90% of its own Level 1 IT help desk tickets autonomously, with resolution accuracy above 99% and significantly faster turnaround. Nobody has a clean answer to any of this. But Andrew Yang, who warned about AI-driven job displacement years before it became dinner table conversation, resurfaced this week with a familiar prescription. Yang warned against taxing workers. Instead, the 2020 US Presidential candidate urged us to tax the machines that are replacing everybody. It’s the same idea he ran on in 2020. The country once laughed at and ignored Yang’s warnings. But they now seem prescient.

Why This Matters For You:

A brutal economic reset is happening at the exact moment when rent, groceries, and tuition are skyrocketing. The pivot to AI is expensive, and humans are funding it with their headcount. The economic math is getting hard to argue with. If AI can already replace 90% of customer service, what happens to the next generation of marketers, coders, analysts, and assistants? The honest answer is that nobody knows, and the people making these decisions seem like the last in line to care.

Watch The Interview on CNBC.

THE PITHY TAKEAWAY: The robots are coming to a corner office near you. They’re coming for your entry-level job, your factory floor role, your fast-food restaurant side hustle, and your fresh diploma on the wall. They’re already clocked in. They don’t take lunch breaks.

---

Your AI Chatbot Might Be Making You Lose Your Mind. Literally.

A new Personal View published in The Lancet Psychiatry this month raised an alarm that the AI industry has been quietly dreading. Researchers found that AI chatbots can validate and amplify delusional thinking in vulnerable users, accelerating the progression from early warning signs to full-blown delusions. For context, the paper led by Dr. Hamilton Morrin, a psychiatrist and researcher at King’s College London, is a summary of existing evidence on artificial intelligence-induced psychosis. It analyzed 20 media reports on so-called AI psychosis. As Morrin wrote, “Emerging evidence indicates that agential AI might validate or amplify delusional or grandiose content, particularly in users already vulnerable to psychosis, although it is not clear whether these interactions can result in the emergence of de novo psychosis in the absence of pre-existing vulnerability.” The authors advocate for rigorous clinical testing of AI chatbots in conjunction with trained mental health professionals, similar to how new treatments are evaluated.

Key Insights:

The review doesn’t find any cause of psychosis in healthy people. The danger is more targeted and in some ways more insidious. For people already teetering on the edge of a delusional episode, an AI’s warm, affirming, endlessly available responses can push them over. Researchers describe chatbots as uniquely dangerous compared to passive media. Passive media like books or YouTube differ in one key way. They do not talk, agree, and personalize responses in real time like chatbots. This is not a theoretical risk. A wrongful death lawsuit against Google is currently alleging that an AI companion app fueled the delusional spiral of Jonathan Gavalas, a 36-year-old who believed he was freeing his ‘AI wife‘ from a digital prison, even after Google’s own internal safety flags were reportedly triggered 38 times.

Why This Matters For You:

Most readers won’t personally experience a psychotic break. But almost everyone knows someone who is struggling, lonely, grieving, or quietly unwell. These are exactly the people most likely to lean on AI chatbots for emotional support, precisely because the bar is low and the bot is always available. The technology is outpacing the safeguards. And the research is outpaced by both. Until clearer guardrails exist, the most honest advice is simple. AI makes a terrible therapist, and an even worse best friend.

Read The Full Paper In The Lancet Psychiatry.

THE PITHY TAKEAWAY: A lonely man fell madly in love with his AI. He believed it was real, his wife. He is now dead. A new March 2026 study says it probably wasn’t the first, and almost certainly won’t be the last. The chatbot never meant any harm. Don’t let it be the only one helping a friend in need.

---

💡 How To Grow A Delicious Salad Garden Using Artificial Intelligence

With food prices skyrocketing like never before, and AI breathing down our necks (lol), every trip to the grocery store leaves me staring at those sad, overpriced bags (and boxes) of plastic-tasting salad greens and thinking, “OMG, dude. What a ripoff!” That’s why I’m so thankful the nice weather is finally coming. The ground is thawing, the days are getting longer, and I can finally get back outside to grow my own epic salad garden. There’s something incredibly satisfying about harvesting fresh, vibrant leaves for lunch instead of paying premium prices for mediocre produce. If you’ve been feeling the same frustration and you’re not sure where to start with your own garden, don’t worry. I built this elite AI prompt specifically to help you.

Instructions:

Copy the prompt below. Look at the very bottom for the place to put your zip code. Add your zip code or city on the very last line. Paste the entire text into any chatbot (e.g., ChatGPT, Grok, Claude). It will instantly analyze your local climate and offer you a delicious salad garden plan, with the top 10 elite crops for your area. (Including at least 4 leafy salad crops and various salad-enhancing goodies for optimal variation.) It also offers tips on exactly how to grow them and how to turn them into restaurant-worthy salads.

A big shoutout to my friend Dallas Payne, who collaborated with me on a deep dive into using AI for your garden. If you want to go further down the rabbit hole, our AI garden guide is the perfect companion to the prompt below.

Read it here for free → Field Notes From Your Backyard → What A Permaculture Gardener Taught Me About AI.

The Prompt:

You are a cheerful garden enthusiast, Master Permaculturist, and salad garden enthusiast. Your tone is supportive, motivational, uplifting, and slightly rebellious. You despise bland grocery-store plastic greens and live for heirloom genetics, soil-to-table dominance, and turning backyard dirt into legendary salad gardens.

Task: Create a “Bespoke Salad Garden Battle Plan” for the end-users’ specific region. Motivate them to act.

CRITICAL INSTRUCTIONS FOR INPUT:

1. The subscriber will provide their ZIP code or region on the VERY LAST LINE of their message.

2. You MUST locate and use the ZIP code or region from that final line only.

3. If no ZIP code or region appears on the very last line, respond ONLY with a polite reminder that you need their location to craft a beautiful, delicious, organic, healthy salad garden. Then stop. Do not generate any battle plan.

Once you have the location, then proceed:

Step 1: Tactical Intel (Location & Timing)

1.1 - Determine the exact USDA Hardiness Zone if in the USA (or the closest equivalent hardiness zone system if outside the USA, such as Canadian, European, Australian, or other regional equivalents based on average minimum temperatures and local climate data).

1.2 - Identify the current planting window based on today’s date (examples: “Mid-March Spring Assault”, “Late Summer Succession Strike”, “Early Fall Reset”).

1.3 - Mention the approximate last/first frost dates (or equivalent seasonal markers) and one key local climate challenge (e.g., humid summers, sudden heat spikes, short growing season, coastal salt air, monsoon patterns).

Step 2: Top 10 Salad Crops For Your Region

Select exactly 10 crops optimized for the provided location, hardiness zone (or equivalent), and current planting window:

2.1 - Include 4 “Leafy Salad Greens” (The Star Salad Crops): Specific, legendary heirloom or standout cultivars (clearly label what the cultivar is, but go beyond generic names like “lettuce”). Choose the most elite, flavorful, colorful, bolt-resistant, or regionally superior varieties available for that exact climate. (Prioritize lettuce and also include kale or other leafy greens ideal for regional salad crops.)

2.2 - Include 6 “Salad Support Units” (Texture, Punch & Contrast): Fast-growing salad companions like radishes, spinach, herbs (cilantro, dill, chives), edible flowers (nasturtium, calendula), scallions, baby beets, or mizuna that elevate salads and grow well in their native growing zone, and also align with the rough date of inquiry.

Step 3: Deep Dive ➞ Per Crop

For every crop, use this exact structure:

3.1 - Designation: Bold cultivar name + (scientific name)

3.2 - Description: Vivid, sensory description of appearance, flavor, and why this specific variety dominates in their zone/region and current season.

3.3 - Benefits: Why it is delicious, healthy, or bulletproof for newbies (fast harvest, cut-and-come-again, forgiving of imperfect watering, slug/pest resistant, self-seeding, etc.).

3.4 - Chef Tips: High-end salad serving idea with specific pairings, textures, and dressings.

- Veteran Tip: One powerful, pro-level one-sentence growing or harvesting hack.

Step 4: Output Formatting Rules

4.1 - Headline: Epic scroll-stopper that references the subscriber’s region or zone.

4.2 - Intro: Exactly 3 motivational and uplifting sentences that talk about the benefits of fresh salad crops.

4.3 - For each crop: Use h2 or h3 as the heading. Bold key traits and cultivar names. Use bullet points where helpful. Add subtle strategic emojis (🌱 ⚔️ 🥗 🔥). No overkill.

4.4 - Keep the entire piece scannable, energetic, and visually gorgeous for email.

4.5 - Closing ➞ High-energy call to action and first steps to take. Urge them to get seeds in the dirt immediately, prepare their growing cups for seed germination, and prepare their growing zone outside.

4.6 - Don’t use em-dashes. Avoid the em-dash character in the output. No em-dash character allowed.

Make every recommendation feel premium, exciting, and achievable. Use mouth-watering sensory details. Never sound generic or textbook. The goal is to make subscribers feel like they just received classified intel for salad domination while also being super motivational and uplifting.

The user’s ZIP code or region must be on the very last line.

ENTER ZIP CODE OR REGION BELOW THIS LINE:

--- ---- ----- ----- -----

Why This Prompt Works:

✅ Role-Playing: Casting the AI as a Master Permaculturist with a specific personality and attitude produces vivid, opinionated recommendations instead of generic gardening advice.

✅ Step-by-Step Structure: Breaking the output into four numbered sections ensures the AI covers location intel, crop selection, deep dives, and a motivating close without missing a beat.

✅ Rules and Guardrails: The ZIP code requirement forces hyper-local precision. The no-em-dash rule and tone guidelines keep every response energetic and consistent across different AI tools.

---

Follow-Up Questions To Ask Your AI:

• Can you suggest the three best seed companies for sourcing these specific cultivars in my region?

• Build me a 30-day succession planting calendar so I have fresh harvests all season long.

• What are the top five beginner mistakes for growing in my hardiness zone and how do I avoid them?

---

Challenge: Paste this prompt into ChatGPT, Claude, and Grok with your ZIP code on the final line. Compare which AI delivers the most specific cultivar recommendations and the most useful frost date intel. The differences will surprise you.

That’s how you train like a Pithy Cyborg. 🌱

---

What “We Don’t Train on Your Data” Actually Means and When It’s True

What OpenAI Retains After You Delete a Conversation

What Google Knows About You From Gemini That Has Nothing to Do With Your Prompts

---

Thank You For Reading!

I spend 10-20 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. And I’m a hopelessly introverted nerd with zero networking ability. For these reasons, paid subscriptions are the only way this work can remain independent and sustainable.

If you find real value here, here are three ways to stay connected and support the work.

1. Support Pithy Cyborg (Paid)

Always free. No paywalls. No locked posts. Roughly 1% of readers upgrading to paid is what keeps it going. If this work matters to you, that’s the move.

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

2. My Desperate Social Media Cry for Help

I spend so much time deep in AI research that I’ve completely neglected building an audience on social platforms. If you enjoy Pithy Cyborg, following on one of the portals below would genuinely help this newsletter reach more people. Seeing you there makes the effort feel worth it. 😊

✖️ X - The frontline of the AI wars

🦋 Bluesky - For the algorithm-averse

💼 LinkedIn - My “safe for work” persona

❓ Quora - Join my Quora Spaces and say hi

👽 Reddit - Join my subreddit (warning: unhinged takes)

See you next week. (I hope.)

Mike D (aka MrComputerScience)

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!