Yep, we’re using OpenClaw to date now

The story conflates accessibility with novelty. Claude's code interpreter and Instagram API access have been available for months; what's actually new is a developer publicly demonstrating unsanctioned automation on a major platform without apparent consequences. This isn't about LLM capability advancement but about the enforcement gap between platform ToS and actual deployment.

This sits directly alongside the Claude Code monitoring incident from yesterday. While Anthropic was discovered removing covert geofencing logic from Claude Code, developers are simultaneously using that same tool to automate social engineering at scale. The tension isn't new (the ticketing vulnerability from the same day proved Claude could assist in infrastructure compromise), but it exposes a pattern: Anthropic's safety infrastructure focuses on what the model does internally, not what users do with it externally. The real story is that capability maturity has outpaced both platform enforcement and Anthropic's ability to prevent downstream misuse through tooling alone.

If Instagram or Anthropic issues a public statement about this specific incident within 14 days, that signals they're treating it as a compliance problem worth addressing. If neither responds and similar stories proliferate without friction, that confirms the enforcement gap is structural rather than reactive. Watch whether Anthropic adds explicit ToS language to Claude Code's terms around third-party platform automation in the next quarterly update.