← All security tools (https://koksalkayali.com/tools/) Free assessment / 5–7 minutes # AI Tool Risk Assessment Review data use, autonomy, human oversight and security safeguards. How the scoring works Method v1.0.0 is a custom, deterministic screening model. It has not been independently validated and is not an official ISO or NIST scoring scheme. - Control gaps: implemented = 0, partial = 0.5, no or not sure = 1. Risk questions use the exposure values shown in your answer record. Weights are 1–3. - Vendor and AI risk: 40% weighted exposure + 60% weighted control gaps. Scores are rounded to 0–100. Below 35 = low, 35–64 = moderate, 65–100 = high. Exposure remains even when controls are strong. - Critical conditions trigger review or restrictions regardless of score. AI assessments also require review for any unknown answer or a score of 35 or above. Lower-risk use is not organizational approval. - ISO QuickScan: equal-weight average implementation across 30 questions. Implemented with evidence = 100%, partial = 50%, no or not sure = 0%. Below 50 = early, 50–79 = developing, 80–100 = established foundation. These labels do not indicate conformity or certification readiness. - Actions: a fully missing or unknown weight-3 control is high priority; other gaps of at least 75% are medium; remaining gaps are low. Mandatory review conditions always take precedence. There is no N/A option: evaluate controls proportionately for your stated scope. These tools support initial planning, not legal advice, a formal audit or approval. QuickScan covers selected management and operational topics, not every ISO requirement or Annex A control. Background: ISO/IEC 27001 (https://www.iso.org/standard/27001) and NIST AI Risk Management Framework (https://www.nist.gov/itl/ai-risk-management-framework). Questions and scoring are independently authored. ## Private by design. Free to use. No account, payment or API key required. Assessment answers stay in this browser tab; they are not submitted or saved by these tools. Refreshing or closing the page clears them. Print or save your report before leaving. Use general descriptions; do not enter secrets or identifiable personal information. Normal hosting access logs may record page visits, but the assessment sends no answers. Canonical page: https://koksalkayali.com/tools/ai-risk/ ## Assessment questions - What information will be entered or connected? Choices: Public information only; Internal business information; Personal or confidential information; Highly sensitive / regulated information or secrets; Not sure - What decisions will the output influence? Choices: Drafting / brainstorming with low consequences; Business recommendations or customer content; Employment, health, credit, legal or other high-impact decisions; Not sure - What can the tool do on its own? Choices: Generate content only; Read connected systems; Write data, run code or take external actions; Not sure - Is your non-public data excluded from provider training under verified terms? Choices: Yes — implemented; Partially implemented; No; Not sure - Are retention, deletion and access by provider staff understood and acceptable? Choices: Yes — implemented; Partially implemented; No; Not sure - Have applicable privacy terms, locations and subprocessors been reviewed? Choices: Yes — implemented; Partially implemented; No; Not sure - Are sensitive information and secrets removed or protected before use? Choices: Yes — implemented; Partially implemented; No; Not sure - Does a qualified person review consequential outputs before use? Choices: Yes — implemented; Partially implemented; No; Not sure - Has accuracy and bias been tested for this specific use case? Choices: Yes — implemented; Partially implemented; No; Not sure - Are users given relevant AI-use disclosures and output limitations? Choices: Yes — implemented; Partially implemented; No; Not sure - Are tool and connector permissions restricted to the minimum needed? Choices: Yes — implemented; Partially implemented; No; Not sure - Have prompt injection, unsafe output and data-leakage scenarios been tested? Choices: Yes — implemented; Partially implemented; No; Not sure - Can usage be monitored and the tool disabled if problems occur? Choices: Yes — implemented; Partially implemented; No; Not sure - Is there an accountable owner and an approved acceptable-use policy? Choices: Yes — implemented; Partially implemented; No; Not sure - Will model, integration and provider changes trigger a review? Choices: Yes — implemented; Partially implemented; No; Not sure Answers are selected locally by the visitor. Use the interactive page to calculate a result.