Posit AI policy puts human review and disclosure in focus
The October policy outlines responsibilities for researchers and analysts using AI assisted work.
Posit’s AI Pass Responsible Use Policy, dated October 1, places human judgment and research integrity at the center of its AI service. It asks users to verify code, analysis and conclusions before relying on them, and addresses disclosure when AI contributes materially to work intended for others.
The policy prohibits presenting AI output as human generated and omitting disclosures required by a publisher, institution, professional standard or law. It also bars consequential decisions about people without meaningful human review. Posit says the policy supplements its service terms and acceptable use rules.
The service terms add important boundaries
The accompanying service terms contain a separate restriction on using output about a person for decisions with legal or material effects, including employment, education, healthcare and credit. They also prohibit entering highly sensitive information such as health data, financial information and government identification numbers.
The terms say users must assess the accuracy and suitability of output before using or sharing it. Researchers considering a workflow should therefore read the relevant documents together and check their institution’s own requirements before putting a consequential process into service.
Check the retention terms for the selected model
The service terms describe zero data retention agreements for certain providers and models, rather than every model. They direct users to product documentation for the current coverage. They also describe an option governing whether content can be used to help train Posit’s models.
That distinction matters when choosing a tool for a sensitive project. A broad commitment on a policy page should not replace checking the actual model, account setting and permitted data for the workflow. Teams can record those choices alongside the version of the policy they reviewed.
Make review part of the workflow
A useful implementation is to name the person responsible for checking generated analysis before it reaches a report or publication. The review can cover source data, executable code, citations and the description of AI’s role. That makes responsibility visible at the point where a draft becomes work other people may rely on.
Featured image is an original AI generated conceptual editorial illustration.
