Most people being laid off for AI aren't being replaced by AI.

That's not my opinion. That's what the data now says — and what one of the most powerful people in Silicon Valley just confirmed out loud. I've been sitting with this for two days and I can't stop thinking about it.

Here's what I mean.

Here's what most people believe: AI is getting so good that companies have no choice but to cut staff. The narrative is clean. Technology advances. Jobs disappear. Move on.

But this week, Marc Andreessen said the quiet part out loud on the 20VC podcast. His exact words: "Every large company is overstaffed by at least 25%. Most are overstaffed by 50%. A lot of them are overstaffed by 75% — and now they all have the silver bullet excuse."

That excuse is AI.

Here's the part nobody in the coverage picked up. A Forrester study found that 55% of employers who laid people off "because of AI" already regret it. Two out of three are quietly rehiring. And one in three spent more on restaffing than they saved from the cuts.

Everyone says AI is replacing workers. The data says companies are firing people they already wanted to fire — and blaming a technology that hasn't even delivered on its promises yet.

I'll be honest. This doesn't make me feel better. It makes me feel worse. Because if the layoffs aren't really about capability — they're about convenience. And that means the people getting cut aren't behind. They're just unlucky.

What to do with this: Open ChatGPT or Claude right now and type: "I work as a [your role]. What are 3 ways I can document my impact so my contributions are visible to leadership?" Takes 4 minutes. Do it before your next meeting.

The Signal: The biggest career threat in 2026 isn't AI replacing you. It's being invisible when your company decides who stays.

So many of you reply saying "I know I should be using AI but I don't know where to start." I hear you. And that's exactly why I do these calls.

In one session I'll audit your workflow, find the biggest time drains, and we'll build automations together using your tools. Not theory. Not "here's a course." We sit down, we set things up, and you walk away with systems already running.

I've done this with 100s of professionals now and every call looks different because every person's work is different. That's the whole point.

→ Book your session

You know that feeling when you search something personal — a health question, a financial worry, a career fear — and you use AI instead of Google because it feels safer?

A class-action lawsuit filed this week in San Francisco says Perplexity has been downloading trackers onto your device the moment you log in. Those trackers gave Meta and Google full access to your conversations. Even in Incognito mode.

The lawsuit alleges Meta and Google used that data for ad targeting and resold it to third parties. Perplexity's response? "We have not been served any lawsuit that matches this description."

The twist: people switched to AI search specifically to avoid being tracked by Google. Now the data went straight back to Google anyway.

What to do with this: Check what accounts and trackers are connected to your AI tools. Perplexity's privacy settings — takes 2 minutes — review what data you're sharing.

The Signal: The AI tool you trust with your private thoughts might be the one selling them. Check your settings today.

A debugging file was accidentally bundled into a routine update of Anthropic's coding tool, Claude Code. It contained nearly 2,000 files and 500,000 lines of source code — pushed straight to a public registry.

Within hours, the entire codebase was mirrored on GitHub and had thousands of stars.

No customer data was exposed. But the leak revealed unreleased features, internal performance data, and a "persistent assistant" running in background mode that nobody knew about.

The company that positions itself as the responsible AI lab made a packaging error a junior developer would catch.

What to do with this: Not a reason to panic. But a reason to ask: what's running in the background of your AI tools? Review the permissions of any AI tool with access to your files.

The Signal: Safety isn't a brand. It's a practice. And even the best slip up when moving fast.

You keep hearing me talk about AI agents. So let me tell you what I actually use.

I've been recommending Lindy AI for months now because it's the easiest way to get started if you've never built anything before. No code. No tech skills. You just describe what you want in normal English and it sets up the whole thing.

I have one that reads my inbox every morning, flags what matters, and drafts replies in my voice. Another one pulls new leads into a spreadsheet automatically. Took me maybe 10 minutes each to set up and I haven't touched those tasks manually since.

If you've been reading this newsletter thinking "OK but how do I actually start using AI for real" then this is your answer. Pick one boring task you do every week. Automate it. You'll wonder why you waited so long.

→ Use Lindy AI