When you use Superhuman’s generative AI assistance, you may occasionally see an error message letting you know that it can’t help with your prompt:
Can’t help with this action
Assistance is unavailable for this action because of security guardrails.
Why this happens
Superhuman’s generative AI has built-in safety guardrails that prevent potentially harmful content from being processed. When something in your text or document appears to trigger one of these guardrails, no output is generated, and you’ll see this message.
In some cases, a document may contain content that is hidden or added without your knowledge and designed to make an AI system behave in unintended ways. Our protections are built to catch this kind of content before it’s processed, helping keep you and your data safe.
Note: Because these protections are designed to err on the side of caution, they may occasionally flag text that is completely harmless. We’re continually working to improve their accuracy, and we appreciate your patience as we do.
What you can check
If you run into this error message, here are a few things that may help you identify what triggered it:
- Rephrase any text that reads like instructions. Wording that sounds like a command or set of directions to a computer or AI system can sometimes trigger these guardrails. Try rephrasing or removing it, then submit your text again.
- Review any text or documents from an external source. If you received the document from someone else, downloaded it, or copied it from the web, review it carefully. Content from outside sources is more likely to contain material you didn’t add yourself.
- Look for hidden or hard-to-see content. Check for text in a very small or white font, unusual formatting, or anything that isn’t visible at first glance. These are common ways that unwanted instructions can be concealed within a document.
If you believe your text was flagged in error
If you’ve reviewed your text and believe it was flagged by mistake, please contact Support and provide the sample of text that caused this message to appear. If that text is part of a larger document, please share the complete document as well.
We can’t investigate the issue without seeing the text that triggered the message. While we may not be able to resolve it right away, this information helps us continually improve.
To learn more about our company’s stance and commitment to using generative AI in an ethical and responsible manner, see the following articles: