Lately, there’s been a massive trend of people trying to use AI agents (like Claude’s Computer Use, Codex, etc.) to run social media accounts. Unsurprisingly, Instagram is cracking down hard on this, and we are seeing a huge wave of action blocks and permanent bans.
As marketers, the desire to automate repetitive tasks is totally understandable—the efficiency gains are undeniable. However, the risks are higher than ever.
My team manages multiple SNS accounts and we’ve been running tests to understand exactly where Instagram's anti-bot system draws the line. I’m not sharing this to encourage everyone to go out and automate their accounts (you run a high risk of losing them), but rather to share technical observations on how IG's security currently behaves.
If anyone else has hit an action block recently, I’d love to hear what triggered it for you. Here is what we found triggers the algorithm:
- Profile Scraping / Data Collection is surprisingly lenient: Unlike LinkedIn, which will ban you almost instantly for automated scraping, Instagram seems less sensitive here. We observed that collecting data from around 30 to 40 accounts a day using an agent doesn't seem to trigger their security systems.
- Engagement (Likes & Replies) is highly restricted: This is the quickest way to get caught. Based on our tests, the absolute hard limit is 5 consecutive likes or 3 consecutive replies. Pushing an agent past that in a short timeframe triggers an immediate 24-hour action block. Repeated offenses quickly turn into a permanent ban.
- Automated Posting is strictly capped: Attempting to use an agent to push multiple automated posts in a single day flags the account almost immediately. It seems that keeping it to a maximum of 1 automated post per day is the only way to avoid the spam filter.
- The biggest variable is the "Trust Score": Ultimately, how aggressively these limits are enforced depends entirely on the account's trust score (account age, follower count, and history of organic, manual activity). If you try to connect an AI agent to a fresh account, it will likely be banned instantly. From our observations, if an account is going to use any form of AI assistance, it absolutely requires a long period of manual, human operation first to build up that baseline trust.
Has anyone else been testing the boundaries of IG's current anti-bot systems? What specific actions got your accounts flagged? Let’s discuss.