AI misunderstood a simple instruction, then contradicted itself. What happens when an AI agent turns misinterpretation into action?

Description and use

<p>Thanks for support..<br> Release: "Sentinel-IR-MCP v0.2.2 on GitLab" (<a href="https://gitlab.com/jakubrojicek11/sentinel-ir-mcp/-/tree/v0.2.2?ref_type=tags" rel="noopener noreferrer">https://gitlab.com/jakubrojicek11/sentinel-ir-mcp/-/tree/v0.2.2?ref_type=tags</a>)</p> <p>Package: "<a href="mailto:[email protected]">[email protected]</a> on npm" (<a href="https://www.npmjs.com/package/sentinel-ir-mcp/v/0.2.2" rel="noopener noreferrer">https://www.npmjs.com/package/sentinel-ir-mcp/v/0.2.2</a>)</p> <p>AI misread: “I don’t want to save this permanently” as “Save it permanently / forge

Problem

<p>Thanks for support..<br> Release: "Sentinel-IR-MCP v0.2.2 on GitLab" (<a href="https://gitlab.com/jakubrojicek11/sentinel-ir-mcp/-/tree/v0.2.2?ref_type=tags" rel="noopener noreferrer">https://gitlab.com/jakubrojicek11/sentinel-ir-mcp/-/tree/v0.2.2?ref_type=tags</a>)</p> <p>Package: "<a href="mailto:[email protected]">[email protected]</a> on npm" (<a href="https://www.npmjs.com/package/sentinel-ir-mcp/v/0.2.2" rel="noopener noreferrer">https://www.npmjs.com/package/sentinel-ir-mcp/v/0.2.2</a>)</p> <p>AI misread: “I don’t want to save this permanently” as “Save it permanently / forge

Outcome

<p>Thanks for support..<br> Release: "Sentinel-IR-MCP v0.2.2 on GitLab" (<a href="https://gitlab.com/jakubrojicek11/sentinel-ir-mcp/-/tree/v0.2.2?ref_type=tags" rel="noopener noreferrer">https://gitlab.com/jakubrojicek11/sentinel-ir-mcp/-/tree/v0.2.2?ref_type=tags</a>)</p> <p>Package: "<a href="mailto:[email protected]">[email protected]</a> on npm" (<a href="https://www.npmjs.com/package/sentinel-ir-mcp/v/0.2.2" rel="noopener noreferrer">https://www.npmjs.com/package/sentinel-ir-mcp/v/0.2.2</a>)</p> <p>AI misread: “I don’t want to save this permanently” as “Save it permanently / forge

Evidence

AI misunderstood a simple instruction, then contradicted itself. What happens when an AI agent turns misinterpretation into action?

Source