Chrome’s on-device AI APIs expect you to declare the output language everywhere you touch the model; otherwise the browser logs that warning.
My fix was to thread the language code (for example 'en', 'es', or 'ja') through both the availability check and the session creation call before you issue summarize() or rewrite():
```
const OUTPUT_LANG = 'en'; // or 'es', 'ja'
await chrome.ai.summarizer.availability({ outputLanguage: OUTPUT_LANG });
const summarizer = await chrome.ai.summarizer.create({ outputLanguage: OUTPUT_LANG });
await chrome.ai.rewriter.availability({ outputLanguage: OUTPUT_LANG });
const rewriter = await chrome.ai.rewriter.create({
outputLanguage: OUTPUT_LANG,
sharedContext: '…'
});
```
If you skip the language in either place, Chrome assumes “auto” and fires the warning.
Passing the same outputLanguage value in both calls and subsequent API call-satisfies the safety attestation requirement and the warning goes away.