I Got Banned for Automation. Here's What I Learned.

I got my first warning from Meta on a Tuesday afternoon while I was eating lunch. I wasn't doing anything dramatic. I'd been using Claude to automate some of my posting on Threads, the kind of thing every growth guide on the internet tells you is just "being consistent." A few posts a day, scheduled in advance, some light engagement. I'd seen accounts doing far more and growing fine.

The warning said my account was showing "inauthentic behavior." I slowed things down, waited a week, started again more carefully. Two weeks later the account was restricted. Not suspended outright, just quietly throttled so that almost nothing I posted reached anyone outside my existing followers. No notification that this had happened. I found out because the numbers dropped off a cliff and I spent three days convinced I'd just gotten bad at this before someone in a forum told me what shadow banning actually looks like.

I appealed. I got an automated response telling me the decision stood. I appealed again. Same response, word for word.

That was the moment I stopped thinking about this as a compliance problem and started thinking about it as something structurally different.


What the rules actually say versus what they mean

Every major platform has terms of service that prohibit automation, third-party tools, artificial engagement, anything that might be interpreted as "inauthentic activity." The language is broad enough to cover almost anything that isn't manually clicking every button yourself in real time.

What's interesting is that the platforms themselves run entirely on automation. The algorithm deciding which posts to show you is automated. The system deciding how much reach to give your content is automated. The ads served to your audience are automated and the platform profits from them. The moderation system that flagged my account ran on automation. None of that is considered inauthentic. The inauthenticity is only ever on your side.

Meta gives you the API. Then bans you for using it. That's not a bug in their thinking, that's the logic of the whole arrangement made visible.

I'm not saying what I did was without risk, I knew I was in a grey area and made a calculation. What I'm saying is there's something worth examining in who gets to automate and who doesn't. The platform automates everything in service of keeping you engaged and showing you ads. You automate anything and you're threatening the integrity of the community.

Once I saw that framing I couldn't unsee it.


The thing about reach that took me too long to understand

The platform doesn't want you to have a direct relationship with your audience. It wants to sit in the middle of that relationship and charge a toll.

When you build a following on Instagram or Threads, you don't actually have access to those people. You have permission to try to reach them, permission the platform grants and revokes based on factors they don't fully disclose, that change without notice, and that ultimately serve their revenue model rather than your ability to communicate. The follower count is a number on their servers. The reach is a variable they control. The relationship is mediated entirely by a system you have no real insight into.

Automation felt like a shortcut. What I understand now is that it was an attempt to work around a system that was already working against me, and the reason it felt necessary was the system itself. I wasn't fighting the algorithm. I was fighting the business model.


What getting restricted actually felt like

The practical part was bad enough. Months of work, and then watching the numbers quietly collapse while I tried to figure out what had happened and whether it was fixable.

But the stranger part was the feeling of having no one to talk to about it. Not in a support sense, I knew support would be useless. I mean there was no real account of what had happened, no way to understand the decision, no pathway to meaningfully contest it. I had done something the platform didn't like. The platform had responded. That was the entire relationship. The content I'd made, the people who'd followed because they genuinely liked it, the time, none of it entered the equation.

The people I talked to who'd been through similar things all described the same feeling across very different situations. A shop owner whose account was flagged for IP violations on products she'd legally purchased. A creator whose finance channel got demonetized because the algorithm decided the genre was a liability. The specifics were different. The feeling of being a variable in someone else's equation was identical.


The structural problem

These are advertising businesses. The product they sell is attention, and the inventory they sell against is your content and your presence on the platform. You create what keeps people there. The platform sells ads against that attention. You get reach, sometimes, when it serves the platform's interest to give it to you.

That's not a corruption of some original good intention. That's the model. And within that model, your ability to reach the people who follow you is not a right or even a consistent privilege. It's a marketing variable.

Automation threatened that model because it introduced a layer I controlled. I could decide when to engage, with whom, at what scale. That's the part that was inauthentic, not because it harmed anyone, but because it moved a small amount of control from the platform to me.


What I found when I actually started looking

After losing the account I started looking for somewhere that doesn't have this structural problem built in. Not better rules or more transparent enforcement, but something where the architecture itself doesn't put an advertising company in the middle of your relationship with your audience.

That's how I found Nostr, and through it, nostr.blog.

Nostr isn't a platform. It's a protocol, the way email is a protocol. Nobody owns it. Your identity is a cryptographic key that belongs to you, your content lives on relays you choose, and no company can wake up one morning and decide your account violates their guidelines, because there's no central company making those calls. On Nostr, automation isn't a violation. It's a feature. You can connect any AI agent and let it write, schedule, and publish for you, and nobody is going to shadow ban you for it.

Nostr.blog is where I landed because it's the least painful entry point into all of this. You pick a name, pay with Lightning, and have a full Nostr identity in a couple of minutes. No hex strings, no confusing setup. The interface has things I didn't expect to find useful and then couldn't imagine not having, a relay manager that shows you real-time ping so you can see which relays are fast and which are lagging, a media library organized like a file system, zap analytics that show you what percentage of your posts actually earned sats. There's a built-in wallet via NWC so you can see your balance without leaving the dashboard. There's even a lo-fi music player with 55 tracks for when you're writing.

None of that is the point, exactly. The point is that when I publish something there, I own it. Not in the terms-of-service sense where a company pinky promises not to delete your stuff. In the architectural sense where the content exists on multiple relays I've chosen, signed with my key, and no single entity can make it disappear. My audience follows me across every Nostr client automatically. If nostr.blog disappeared tomorrow, I'd open any other Nostr client, log in with the same keys, and everything would be right there.


What I'd tell someone starting now

Don't automate on Meta, not because the rules say so but because you'll spend months optimizing for a number the platform controls and can change without telling you.

But more than that: be honest about what you're actually building and where. If the answer is "an audience on a platform that needs that audience to sell ads," you're not building something you own. You're contributing to something someone else owns, and your access to it is conditional in ways you probably haven't fully mapped out.

That's not necessarily a reason not to be there. There are real people on these platforms and real reasons to show up. But knowing what the actual arrangement is changes how you think about where else your work should exist.

I wished I'd thought about that before I was sitting watching everything I'd built go quiet on a Tuesday afternoon.


AI Automation for Freelancers: How to Earn More While Working Less
How freelancers can use AI automation to earn more, work fewer hours, streamline workflows, and grow their business with smarter, faster systems.
How to Automate Your Workflow with AI Agents: Complete 2026 Guide
The workflow automation landscape transformed dramatically in 2026 with the rise of agentic AI—intelligent systems that don’t just respond to commands but autonomously complete multi-step tasks across your entire tech stack.
The Quiet Stack Powering Real AI Automation in 2026: n8n, Dify, and Ollama
How n8n, Dify, and Ollama became the go-to open-source stack for teams building private, production-grade AI automation in 2026 — and why it works.