AI Disclaimer

1. Purpose of This Disclaimer

ProofInk.AI uses artificial intelligence to provide feedback on grant proposals and research papers. This page explains what our AI can and cannot do, how your data is handled during AI processing, and the limitations you should be aware of before relying on our output.

2. What Our AI Does

ProofInk.AI analyzes your documents against evaluation criteria and generates structured feedback. Specifically, our AI:

• Evaluates your document against agency-specific or discipline-specific criteria using calibrated rubrics
• Identifies strengths and weaknesses with mandatory citations from your document text
• Detects gaps, contradictions, and inconsistencies across sections
• Generates scores calibrated to concrete rubric anchors
• Produces a prioritized improvement roadmap with specific, actionable suggestions
• Performs cross-validation to verify internal consistency of its own findings

The analysis is generated by large language models (currently from Anthropic and/or OpenAI) that have been configured with domain-specific evaluation frameworks developed by our team.

3. What Our AI Does NOT Do

• It does not replicate the judgment of a human expert reviewer or review panel
• It does not guarantee that addressing its suggestions will result in a funded proposal or accepted paper
• It does not provide official feedback from any funding agency, journal, or publisher
• It does not verify the scientific accuracy or validity of your research claims
• It does not replace peer review, mentor feedback, or institutional review
• It does not have access to information about your track record, institutional context, or the broader funding landscape
• It does not know the composition or preferences of the specific review panel that will evaluate your work

4. Known Limitations

• Imperfect analysis: like all large language models, our AI may occasionally generate feedback that is inaccurate, incomplete, or not fully applicable to your specific research context. Always verify AI-generated feedback against your own expertise.
• Field specificity: while our system adapts to different research fields, it may not capture highly specialized disciplinary norms, emerging methodological standards, or niche subfield expectations.
• Context limitations: the AI evaluates only the document you upload. It cannot assess external factors such as your publication record, preliminary data not included in the document, team expertise, or institutional resources.
• Score calibration: scores are calibrated against general evaluation rubrics derived from publicly available criteria. They are not calibrated against the specific scoring behavior of any particular review panel or agency program officer.
• Document quality: the quality of AI feedback depends on the quality and completeness of the uploaded document. Poorly formatted, incomplete, or very short documents may yield less useful results.

5. AI Service Providers

Our AI analysis is currently powered by models from Anthropic (Claude) and/or OpenAI, accessed via their APIs. Key commitments from our AI providers:

• Data sent via the API is not used for model training
• Data is not retained for model training and is automatically deleted after a short safety-monitoring window
• Processing occurs in the United States

For more information, see Anthropic's policies and OpenAI's enterprise privacy.

We may change AI providers in the future. Any new provider will be held to equivalent or stricter data protection standards, and this page will be updated accordingly.

6. Data Handling During AI Processing

When you run a feedback report, the text content of your uploaded document is sent to our AI provider's API for analysis. The following safeguards are in place:

• Document content is transmitted over encrypted connections (HTTPS/TLS)
• Backups containing your documents are encrypted with AES-256 before leaving our servers
• Content is not retained by the AI provider for model training and is automatically deleted after a short safety-monitoring window
• Content is not used to train, fine-tune, or improve the AI provider's models
• Content is not accessible to the AI provider's employees except as required for security monitoring
• Input content is sanitized before processing to neutralize potential prompt injection attempts

7. Agency Disclaimer

ProofInk.AI is not affiliated with, endorsed by, or sponsored by any funding agency or research council, including NSF, NIH, DOE, Simons Foundation (USA), NSERC, CIHR, SSHRC (Canada), ERC (EU), UKRI, Wellcome Trust, NIHR (UK), or any other organization. Review templates and evaluation criteria are based on publicly available solicitation documents and general grant-writing best practices — they do not constitute official agency guidance. All agency names, acronyms, and trademarks are the property of their respective organizations and are used solely for identification purposes.

8. Your Responsibility

By using ProofInk.AI, you acknowledge that:

• You are solely responsible for all decisions regarding your research, grant applications, and manuscript submissions
• AI-generated feedback is informational only and carries no guarantee of accuracy or completeness
• ProofInk.AI assumes no liability for outcomes resulting from your use of our service
• You will exercise your own professional judgment when interpreting and acting on AI-generated feedback

9. Academic Integrity and AI Disclosure

ProofInk.AI is a feedback tool — it reviews and critiques your work, it does not write or rewrite your documents. Using ProofInk.AI is comparable to receiving feedback from a colleague or writing center: the intellectual content remains entirely yours.

However, institutional policies on AI tool usage vary. Some universities, funding agencies, and journals require disclosure of AI-assisted tools used during the preparation of a submission. We recommend that you:

• Check your institution's policy on AI tool usage before using ProofInk.AI for work subject to academic integrity requirements
• Consult your funding agency's guidance on AI-assisted preparation of proposals (e.g., NSF, NIH, ERC, UKRI, and NSERC each have evolving policies)
• Review your target journal's AI disclosure requirements if using ProofInk.AI for manuscript preparation
• When in doubt, disclose that you used an AI-based feedback tool to review your draft — transparency is always the safest approach

ProofInk.AI does not generate text for inclusion in your documents. It analyzes what you have written and provides structured feedback. The final document you submit is your own work and your own responsibility.

10. Document Originality and Isolation

Your documents are stored in isolated, user-specific storage. No other user can access, view, or search your files. Documents are never indexed, published, shared with third parties, or included in any database accessible to others — including plagiarism detection services, search engines, or content aggregators. The originality and confidentiality of your academic work is fully preserved throughout the feedback process.