Progress 0 / 22 questions answered

IEEE 7000-series Ethics Checklist

Assess your AI system's ethical compliance. 22 questions, 5 categories.

Values-Based Design & Stakeholder Engagement

IEEE 7000 standard: Human values considered during AI system design and stakeholder involvement.

Have you identified human values affected by the AI system (e.g., fairness, autonomy, safety)?

IEEE 7000: Values elicitation. AI is not value-neutral – every decision reflects values. These must be chosen consciously.

No, not considered Discussed informally Documented via a value elicitation workshop

Have you engaged stakeholders in the AI system design process?

IEEE 7000: Stakeholder engagement. Input from stakeholders (users, affected parties) is critical for ethical AI.

No, only developers decided Conducted a user survey Stakeholder workshops and participatory design

Have you addressed value conflicts (e.g., privacy vs. safety, fairness vs. accuracy)?

IEEE 7000: Value trade-offs. AI design often involves value dilemmas. Decisions must be made explicit.

No, not addressed Implicit decisions (not documented) Documented trade-off analysis and value ranking

Is there an ethics review process in place for AI projects?

IEEE 7000: Ethics review. Similar to an IRB (Institutional Review Board) in research. Independent review is required.

No ethics review Internal review (team level) Independent ethics board/committee review

Do you monitor the AI system's value alignment after deployment?

IEEE 7000: Continuous value verification. Values are not static – system drift must be monitored.

No post-deployment monitoring Value-aligned metrics tracking Regular ethics audit and stakeholder feedback loop

Transparency & Explainability

IEEE 7001 standard: Transparency of AI system operations and decisions.

Do you clearly disclose to users that they are interacting with an AI?

IEEE 7001: AI disclosure. People have the right to know if they are interacting with a machine (not a human). The EU AI Act also requires this.

No, not disclosed Disclosed, but hidden in fine print Clear AI disclosure in the UI (e.g., "AI Assistant")

Can the AI's decisions be explained to users?

IEEE 7001 & GDPR Article 22: Right to explanation. "Black box" AI is increasingly unacceptable.

No, it's a black box model Partial (e.g., feature importance / SHAP) Full explainability (e.g., LIME, counterfactuals, attention maps)

Do you have a Model Card or similar documentation for the AI system?

IEEE 7001: Model transparency. A Model Card (Google) or Datasheet (Microsoft) is the AI model's "label" (use case, limitations, training data).

No Model Card Internal documentation exists Public Model Card (Mitchell et al. format)

Is it transparent what data the AI model was trained on?

IEEE 7001: Training data transparency. For AI, "you are what you eat" – the training data determines its behavior.

Secret, not disclosed High-level description (e.g., "web text") Detailed datasheet (sources, sizes, bias analysis)

Data Privacy & User Rights

IEEE 7002 standard: Personal data protection in AI systems.

Do you apply the data minimization principle (collecting only necessary data)?

IEEE 7002 & GDPR Article 5(1)(c): Data minimization. The less data you collect, the lower the risk.

No, "collect everything" approach We filter out some unnecessary data Strict data minimization audit and documentation

Is there clear consent management for AI data usage?

IEEE 7002 & GDPR Article 7: Consent management. Users must know how their data is being used.

No consent, or only generic ToS Consent exists, but it is not AI-specific Granular, AI-specific consent (purpose limitation)

Do you apply Privacy by Design principles during AI development?

IEEE 7002 & GDPR Article 25: Privacy by Design. Privacy is not an afterthought "add-on", but a core, built-in principle.

No, privacy was added as an afterthought Some privacy tech used (e.g., encryption) Full PbD (anonymization, differential privacy, federated learning)

Do you ensure data subject rights (e.g., access, rectification, erasure, portability)?

IEEE 7002 & GDPR Chapter III: Data subject rights. These rights must also be functional in the context of AI.

No, not implemented Manual process (e.g., support ticket) Automated self-service portal

Have you conducted a Data Protection Impact Assessment (DPIA) for the AI system?

IEEE 7002 & GDPR Article 35: DPIA. A DPIA is mandatory for high-risk AI systems.

No DPIA conducted Informal risk assessment Formal DPIA documentation (GDPR compliance)

Algorithmic Bias & Fairness

IEEE 7003 standard: AI system fairness and bias management.

Have you tested the AI system for bias (e.g., prejudice, discrimination)?

IEEE 7003: Bias testing. AI often reflects (or amplifies) societal biases present in the training data.

No bias testing conducted Manual review (based on a few examples) Systematic bias audit (e.g., Fairlearn, AI Fairness 360)

Have you identified protected attributes (e.g., gender, age, ethnicity, religion)?

IEEE 7003: Protected attributes. EU AI Act & GDPR Special Categories. Discrimination based on these is prohibited.

No, not considered A list of protected attributes exists Identified + bias mitigation techniques applied (e.g., reweighting, adversarial debiasing)

Do you measure fairness metrics (e.g., demographic parity, equalized odds)?

IEEE 7003: Fairness metrics. Accuracy is not enough – fairness must be measured separately (but trade-offs exist!).

No, we only measure accuracy One fairness metric Multiple fairness metrics + trade-off analysis

Does your development team's diversity reflect your user base?

IEEE 7003: Team diversity. Homogeneous teams are more likely to have blind spots. Diversity reduces bias.

No, homogeneous team Some diversity Representative diversity + inclusion program

Accountability & AI Governance

IEEE 7000-7003 common themes: Responsibility, audit trail, incident management, and governance structure.

Is there an AI governance structure in the organization (e.g., AI ethics committee, review board)?

IEEE 7000+: AI governance. Ethical decisions cannot be left to developers alone – organizational oversight is needed.

No governance structure Informal review (e.g., manager approval) Formal AI ethics committee or board

Is there an audit trail for AI decisions (who, when, what, why)?

IEEE 7000+: Auditability. AI decisions must be reconstructable (for compliance, incident investigation, etc.).

No logging Basic logging (input/output) Full audit trail (model version, decision rationale, human override)

Is there human oversight for critical AI decisions?

IEEE 7000+ & EU AI Act: Human-in-the-loop. High-risk decisions (e.g., hiring, credit scoring) require a human in the loop.

No, fully automated Human review available on demand (appeal) Mandatory human review for critical decisions

Is there an AI-specific incident response plan (e.g., bias incident, privacy breach, safety failure)?

IEEE 7000+: Incident management. AI incidents (e.g., discriminatory decision) require a fast response – a playbook is needed.

No AI-specific IR plan Generic IT incident plan (not AI-specific) Dedicated AI incident playbook (bias, fairness, safety)

Your AI Security Assessment

/100

📧 Get your results via email

We will send you the complete report with all answers and recommendations

I accept the Terms of Use.

Want a deeper analysis?

Our experts can help with a detailed AI Red Team audit

Schedule Free Consultation →

Is your AI system ethical and responsible?

Take our 3-Minute IEEE 7000 Ethics Checklist! The IEEE 7000 series focuses on designing technology ethically and for social responsibility. This quick checklist helps you assess how well your AI development processes consider transparency, bias mitigation, and human well-being. Complete it in just 3 minutes to get an instant ethics audit score!

We’ll email you the results and recommendations.