I kept telling everyone the Instagram exploit is not "patched" because AI is not linear. The exploit evolves so long as the tooling is still there. I just had one of my OG accounts hit. Got it back but this is June 2, 2026 almost two days post patch.

Forget the Instagram exploit for a minute. Here’s the real story nobody is talking about. AI is here to stay. Every day we trust it with a little more. It automates work, saves time, and in many cases does a pretty good job. The Instagram incident is a glimpse of what’s coming. Whether this specific issue was prompt injection, workflow abuse, or something else entirely almost doesn’t matter. What matters is that we’re putting AI in front of more and more systems and giving it access to more and more data. This won’t be the last time we see something like this. Today it was an Instagram account. Tomorrow it could be your bank account, your smart home, your car, or something else entirely. Every week there are new MCP servers, integrations, agents, and tools being released. Many of them have access to data and permissions that people assume are private and secure. The uncomfortable reality is that large language models are susceptible to manipulation. They can be tricked. They can be influenced. They can be exploited. Prompt injection and prompt exploitation are not theoretical problems. They are real security challenges, and they’re not easily solved. What happened today wasn’t an isolated event. It was a preview.