A newly published, unconfirmed report claims that Google’s Gemini AI has been instructed to observe a user’s tone and ascertain sentiment while basing its responses on fact and reality.
Why do we care. If accurate, AI-generated search answers may vary based on how the query is phrased — not just the information available.
What’s new. The report focuses on the inherent synergy in the system-level commands that guide Gemini’s response. The report, published by Elie Berreby, head of SEO and search AI at Adorama, suggested that Gemini was instructed to:
- Match the user’s tone, energy, and purpose.
- Validate feelings before responding.
- Deliver responses aligned to the user’s perspective.
What does it mean. “Authority that supports too much often exceeds the factual basis,” writes Berreby. So instead of acting as a neutral aggregator, AI responses may:
- Confirm the negative frame (“Why is X bad?”).
- Emphasize the positive frame (“Why is X good?”).
If public opinion is wrong, AI can amplify it. As the report suggests:
- AI reflects existing emotional signals.
- It doesn’t “scale” the way blue links usually do.
Question outline. The emotional frame of the question concerns:
- What sources are cited.
- How to write summaries.
- The perfect tone of the answer.
Google’s AI overview is already showing a shift in tone, which often aligns with the intent of the query beyond keywords. This report provides a possible explanation.
Not guaranteed. Google has yet to confirm the leak. As Berreby noted in his report: “I decided to share with the public only a small part of the leaked internal system information. I am not sharing any sensitive data. This is not a zero-day exploit. This is a small leak.”
Report. This Gemini Leak Means You Can’t Get Over The Feeling
Search Engine Land is owned by Semrush. We are committed to providing the highest quality of marketing articles. Unless otherwise stated, the content of this page is written by an employee or paid contractor of Semrush Inc.
