What Is AI Psychosis? (And Why Your Boss Might Have It)

Last month, Box CEO Aaron Levie posted something that went viral across the entire tech industry. He said that executives are uniquely prone to a condition he called “AI psychosis” — a warped sense of reality where leaders play with AI for five minutes, see the shiny demo, and immediately decide it can replace their entire workforce. Sound familiar? If your boss has been dropping phrases like “AI-first everything” or “agents will handle it” during layoff meetings, you might be living through AI psychosis firsthand. The good news? There is actual data that says this is mostly hype. And knowing the signs can help you protect your career.

—

The Problem: Your Boss Is Hallucinating — And It Is Not the AI

Here is what AI psychosis looks like in real life. Your CEO tries ChatGPT to draft a memo. It works impressively well in about ten seconds. Suddenly, that same CEO is convinced AI agents can write code, manage clients, handle customer support, and basically run the company — all while cutting costs by 60 percent. It sounds absurd when you say it out loud. Yet according to multiple reports, this exact scenario is playing out at companies across the country right now.

In fact, a recent Gartner survey found that roughly 80 percent of companies using autonomous tech have already cut jobs. However, those workforce reductions are not translating into meaningful financial returns. In other words, companies are firing people left and right, but the savings are not showing up on the balance sheet. Meanwhile, nearly 116,000 tech workers have been laid off across 159 companies in just the first five months of 2026 — nearly matching the entire 2025 total in half the time, according to layoffs.fyi.

So why is this happening? The problem comes down to a dangerous combination of optimism, pressure from investors, and a fundamental misunderstanding of what AI can actually do today.

—

What Is AI Psychosis? (And It Is Not What You Think)

Before we go further, it is worth clarifying something. The term “AI psychosis” actually means two completely different things depending on who you ask.

The clinical meaning refers to a documented phenomenon where people develop psychotic symptoms — delusions, paranoia, dissociation — after extended interaction with AI chatbots. Researchers at Harvard and Lancet Psychiatry have identified several types, including cases where users formed deep emotional attachments to AI and struggled to distinguish between the chatbot and reality.

The corporate meaning — which is what we are talking about here — was coined by Aaron Levie on May 24, 2026. Here is exactly what he said on X:

CEOs are “uniquely prone to AI psychosis because they are sufficiently distant from the last mile of work that still has to happen to generate most value with AI.”

Levie is not some AI skeptic, either. He runs a major tech company and actively invests in AI startups. His point is straightforward: executives see the happy path of AI — a prototype that works, a demo that wows — and then leap to believing agents can replace entire jobs. What they miss is the debugging, the edge cases, the training data, the quality review, and the hundreds of human decisions that keep any process running smoothly.

Think of it this way. Anyone can drive a car on an empty road and think they are a great driver. But rush hour traffic, road construction, and a flat tire are what actually test your skills. AI psychosis happens when leaders only see the empty road.

—

Real-World Examples: When AI Psychosis Goes Too Far

ClickUp: The Extreme Case

If you want a textbook example of AI psychosis, look at ClickUp. The project management startup (valued at $4 billion) deployed roughly 3,000 internal AI agents, creating a 3:1 agent-to-human ratio. Then CEO Zeb Evans laid off 22 percent of the workforce, calling it not cost-cutting but a “radical embrace of AI.” Employees now spend their days managing agents instead of doing actual work — reviewing output, fixing errors, and ensuring quality.

Even worse, Evans banned direct messages to the CEO. Employees now have to route everything through an AI agent first. One agent reportedly costs $9 per single run, and they still cannot perform critical tasks like deleting records or pushing code to production. The irony is thick: a company fired humans to rely on systems that cannot do the job fully.

AI-Washing Layoffs: When AI Is the Excuse, Not the Reason

Here is another pattern that has become disturbingly common. Companies announce layoffs and blame AI, even when the AI is not ready to replace anyone. In 2025 alone, AI was cited as the reason for over 50,000 layoffs. Amazon and Pinterest were among the major names.

However, Forrester Research found something telling: “Many companies announcing A.I.-related layoffs do not have mature, vetted A.I. applications ready to fill those roles.” The Brookings Institute went even further, calling AI layoff announcements “a very investor-friendly message.” Forrester now predicts that over half of AI-attributed layoffs will be quietly reversed. In other words, your company may fire you for AI today, then hire someone else to do your job next year.

DuckDuckGo: The Pushback Nobody Expected

It is not just corporate leaders, either. Google recently overhauled Search to force AI Overviews on everyone, with no opt-out option. The result? Users revolted. DuckDuckGo installs surged 30 percent. Visits to noai.duckduckgo.com — a dedicated AI-free search page — jumped 27.7 percent in a single week. CEO Gabriel Weinberg put it plainly: “Google is force-feeding AI with no way to opt out.”

When even everyday users are pushing back against forced AI, maybe it is time for executives to slow down and listen.

—

The Hard Data: AI Is Not Ready to Replace You (Yet)

If your boss has AI psychosis, the most powerful weapon you have is evidence. And the evidence is clear.

MIT FutureTech tested AI agents on over 3,000 workplace tasks. In 2024, they succeeded about 50 percent of the time. By 2025, that improved to roughly 65 percent. At the current pace, researchers estimate agents will reach 80 to 95 percent success by 2029 — but they also note it will take “several additional years” to consistently outperform humans.

Mercor benchmarked top-tier AI models on 480 real workplace tasks. Every single agent failed to complete most duties. Not some — every one.

Berkeley researchers published a major meta-analysis in October 2025 that debunked seven common myths about AI productivity. Their conclusion: “No robust relationship between AI adoption and aggregate productivity gain.”

Nature Human Behaviour reviewed 106 experiments and found something surprising. On average, human-AI teams performed worse than the better of human or AI working alone. The only exception was creative brainstorming tasks.

NBER surveyed 750 corporate executives and found a persistent “productivity paradox” — perceived productivity gains are consistently larger than measured productivity gains. Leaders think AI is doing more than it actually is.

Finally, KPMG found that only 9 percent of companies are actually orchestrating multiple AI agents across workflows. The rest are still in exploration mode. Yet somehow, layoffs are happening as if the revolution already arrived.

—

5 Signs Your Boss Has AI Psychosis

Not sure if your company is just being ambitious or has actually lost touch with reality? Here is a quick checklist.

1. They have seen one demo and now want to “AI everything.” If a single ChatGPT prompt convinced your CEO to restructure the entire department, that is a red flag.

2. Layoffs are framed as “AI transformation” instead of cost-cutting. When companies blame AI for firings but cannot explain what the AI will actually do, that is AI-washing, not strategy.

3. Nobody has measured the current baseline. Before replacing humans with AI, you need to know: How long does the current process take? What is the error rate? What is the cost? If leadership skipped this step, they are guessing.

4. They quote AI vendor marketing as if it is evidence. “Anthropic says agents can do X” is not data. It is a sales pitch. Real evidence means running your own pilots with your own data.

5. The people actually doing the work were not consulted. The most reliable signal of AI psychosis is a top-down mandate with zero input from the employees who understand the day-to-day process.

If three or more of these sound like your workplace, your boss might have AI psychosis. Do not panic — but do prepare.

—

What Should You Do If Your Company Has AI Psychosis?

So your leadership is going all-in on AI and you are caught in the crossfire. Here is a practical survival plan.

Step 1: Become the AI-literate person in the room.

This is the single best career move you can make right now. Learn how AI tools actually work — not at a PhD level, but enough to understand their limitations. If you can explain to your boss why an AI agent cannot handle a specific customer escalation, you become invaluable. For beginners, our beginner guide to Google Gemini Spark is a solid starting point.

Step 2: Document what you actually do.

Most leaders with AI psychosis genuinely do not understand what their employees handle day to day. Start keeping a simple log of your tasks, decisions, edge cases, and the stuff that falls between the cracks. When someone suggests replacing you with AI, you will have a clear picture of what would break.

Step 3: Propose a pilot instead of a revolution.

Instead of fighting the AI push (which rarely works), redirect it. Suggest running a two-week pilot on one specific process, with clear success metrics. This does two things: it shows you are on board with innovation, and it creates evidence. Either the pilot works (great — you helped) or it fails (great — you were right, and now there is data).

Step 4: Keep your options open.

Forrester predicts over half of AI-attributed layoffs will be reversed. That means the job market is going to be volatile. Keep your resume updated, your network warm, and your skills sharp. AI-literate workers are already in high demand, and that trend is only accelerating.

Step 5: Know your rights.

In the US, there are currently no federal protections specifically for AI-related layoffs. However, the WARN Act still requires 60 days notice for large layoffs. If your company is using “AI transformation” as a cover for cuts, document everything. Consult employment resources and, if necessary, legal counsel.

—

The Bottom Line

AI psychosis is real, and it is spreading fast. The term is barely a week old, and it already describes a massive problem across corporate America. Leaders are making billion-dollar decisions based on five-minute demos, while the actual data tells a very different story: AI agents succeed maybe two-thirds of the time on simple tasks, and companies that fire people for AI are not seeing financial returns.

If you are reading this and nodding along, you are not paranoid. You are paying attention. The best thing you can do right now is learn enough about AI to have an informed opinion, document your value, and stay ready. The companies that survive this AI hype cycle will be the ones that use AI to enhance their workforce — not erase it.

—