Responsible AI

• Design, training, and deployment of AI models and agents;

• Integration of third-party AI systems;

• Processing and use of data for AI purposes

• Use of AI outputs by end-users.

• We clearly disclose where and how AI is used within our systems. Users are informed when they interact with AI-driven agents, features, or automated decisions.
  

• We design and test our systems to minimize bias and ensure equitable treatment across all users, irrespective of gender, ethnicity, geography, or language.
  

• We maintain human oversight and responsibility over all AI operations. Decisions made by AI agents are auditable and reversible by authorized personnel.
  

• We actively test AI systems for robustness, misuse resistance, and resilience against malicious manipulation.
  

• We comply with global data protection laws (including PIPEDA, GDPR, CCPA, APPI, PIPA, and PDPA) and ensure that personal data used in AI processes is lawful, minimal, and secure.
  

• AI decisions support — but never override — human intent. AI is built to assist, not dictate.

• OECD Principles on Artificial Intelligence (Promoting inclusive growth, sustainable development, and human-centered values)

• EU Artificial Intelligence Act (2024) (Risk-based classification, transparency, and human oversight)

• Canada’s Artificial Intelligence and Data Act (AIDA) (Responsible innovation and harm prevention)

• U.S. NIST AI Risk Management Framework (Governance, accountability, and measurement of AI risk)

• UNESCO Ethics of AI Recommendation (2021) (Human rights, transparency, and fairness)

Responsibilities

• Evaluate new AI features for ethical and legal risk;

• Approve AI model deployments and major updates;

• Review AI incidents, bias reports, or user complaints;

• Conduct regular audits and publish transparency reports.

• Data Use and Protection (No data is used for AI training, inference, or improvement.)

• The RUNSTACK name, logo, and all related marks are trademarks owned or licensed by RUNSTACK.

• You have no ownership rights in the Platform or its contents.

• Any unauthorized use, reproduction, or modification of our intellectual property may result in civil or criminal penalties.

• Training data is obtained from lawful and ethically sourced datasets. Data is assessed for potential bias, quality, and representativeness.

• We strive for model interpretability and provide documentation that explains how AI agents generate outputs.

• AI systems are tested for safety, bias, accuracy, and resilience before deployment. Performance is continuously monitored to detect anomalies or drift.

• Models are regularly updated to enhance fairness, security, and efficiency.

• Minimal Risk – Chat interfaces, workflow automation, and content summarization.

• Limited Risk – Predictive suggestions, productivity enhancement, or contextual recommendations.

• High Risk – AI-driven decision-support for compliance or security operations (subject to audit).

• Prohibited Uses – AI applications that engage in manipulation, social scoring, surveillance, or rights violations.

• RUNSTACK will immediately suspend the affected system, investigate the cause, and document the incident.

• Affected users will be notified when appropriate.

• Findings will be reported to relevant authorities where legally required.

• RUNSTACK and its affiliates, officers, employees, and partners shall not be liable for any indirect, incidental, consequential, or punitive damages, including lost profits or data, arising from or related to your use of the Platform.

• RUNSTACK’s total cumulative liability to you shall not exceed one hundred (100) U.S. dollars (USD) or the total amount you have paid to RUNSTACK in the past six months, whichever is greater.

• Obtain an explanation for automated decisions that materially affect them;

• Challenge or appeal AI-assisted decisions;

• Opt-out of non-essential AI-driven personalization features;

• Request data access, correction, or deletion; and

• Report ethical or safety concerns to RUNSTACK.

• Requests can be submitted to legal@runstack.ai

• Integrating ethics checkpoints during design and prototyping.

• Embedding compliance and fairness testing before launch.

• Conducting ongoing human-in-the-loop validation.

• Maintaining transparent documentation and user education materials.

• AI system categories in operation;

• Data handling methods;

• Model performance and fairness metrics; and

• Notable improvements to AI governance and safety.