Separate stated facts from unanswered questions
Why now: Restricted rollouts and disputed permissions make precise reading useful before you decide what to adopt or allow.
Prompted by today's lead: Gemini 4 Argon starts with restricted access
What you'll need
- A non-sensitive announcement, feature description or policy excerpt you already have.
Steps
- Choose a short passage and remove any private or confidential information.
- Paste it into the prompt below in a general-purpose AI assistant.
- Compare the answer with the original, especially any statement about access or permission.
The prompt
Read the text below using only what it says. In under 120 words, separate explicit statements, attributed claims and unanswered questions. Focus on availability, eligibility, permissions and evidence of performance where relevant. Do not turn missing information into a negative fact or resolve contradictions yourself. Finish with one concrete question to ask before relying on the announcement. Return plain text with no Markdown formatting, no asterisks and no # headings. Text: [paste your announcement or policy text here]
What you should get
A short distinction between restricted access, the maker’s unverified time-saving claim and missing permission details, ending with a question grounded in the passage rather than an invented answer.
Tested example: Tested on sample input
Input
The new planning assistant is being introduced to selected volunteer groups. Its maker says it reduces time spent organizing shared tasks. Access for other groups will be discussed after the initial review. The announcement says calendar access requires permission, but does not describe the permission screen or explain how access can be withdrawn.
Output
Explicit statements: The planning assistant is being introduced to selected volunteer groups. Access for other groups will be discussed after the initial review. Attributed claims: Its maker says it reduces time spent organizing shared tasks. The announcement says calendar access requires permission. Unanswered questions: How are volunteer groups selected? When will the review occur, and what will determine broader access? What does the permission screen show, and how can calendar access be withdrawn? What evidence supports the claimed time savings? Before relying on the announcement, ask: What measured results from the selected groups support the claim that the assistant reduces organizing time?
How we tested it: How we checked this: an AI model ran this prompt on the sample input on October 1, 2026, and a separate model call checked the result against what the Move promises.