Draft - SOW APPENDIX - 70B03C26Q00000328.pdf
PDF 688 KB Posted
- Attached to
- Global Entry Airport Modernization
- Solicitation number
- 70B03C26Q00000328
About this file
This is a Statement of Work Appendix for the Global Entry Airport Modernization contract (70B03C26Q00000328). The document establishes AI-specific conditions and project management requirements for artificial intelligence implementation within the Department of Homeland Security (DHS).
The AI-specific conditions require contractors to implement safeguards conforming to the NIST AI Risk Management Framework, ensure non-infringement of intellectual property and third-party rights, and address privacy, civil rights, data security, transparency, and fairness concerns. Contractors must assess risks related to unintended disclosure of sensitive data from AI models, submit documentation on data origins and quality, manage biases, and prevent unauthorized data disclosure. The solution must not make decisions based on illegal consideration of protected characteristics including race, ethnicity, gender, national origin, religion, sexual orientation, gender identity, age, nationality, medical condition, or disability in violation of Title VII, the ADA, and the ADEA. For large language models specifically, contractors must provide the original developer's acceptable use policy, model/system/data cards with training summaries and evaluation scores, end-user resources, a feedback mechanism for safety and security violations, descriptions of safeguards for incomplete or contradictory information, content moderation specifications, disclosure of any training conducted outside the United States, disclosure of modifications to comply with non-U.S. government regulations, and controls for modifying outputs if the vendor is not the direct developer.
Project management and documentation requirements include oversight of enterprise data set leveraging, data acquisition, governance, integration, preparation activities, project status updates, risk assessments, technical documentation, business process documentation, and compliance documentation with NIST standards and DHS AI policy.
View the file
Other files
On GovTribe
Work with this file
- Download the original file
- Contacts named in this file
- Similar government files
- Ask GovTribe AI about this file
Text version
Statement of Work Appendix – 70B03C26Q00000328
Global Entry Airport Modernization
Statement of Work (SOW) Appendix
70B03C26Q00000328
AI Specific Conditions
• The contractor shall ensure AI safeguards are in place and employ AI compliance/risk management practices, including ensuring that the AI product or solution conforms to prevalent industry standards and practices (including the NIST AI Risk Management Framework).
• The contractor’s AI product or solution will not infringe upon or violate any intellectual property rights, rights of likeness or publicity, or any other third-party right of any kind.
• The contractor’s AI product or solution shall cover themes such as privacy, civil rights, and civil liberty concerns; ownership and security of data; means to prevent misuse, unauthorized use, or corruption of AI systems; transparency, accountability, and fairness, including data set integrity, accuracy, foreseeable risks, and social impact.
• The contractor shall assess and mitigate risks related to the potential unintended disclosure of sensitive data from AI models, including conducting a model-specific risk analysis before sharing or releasing AI models and model weights.
• The contractor shall submit documentation on the origins, quality, and suitability of the data used to train AI models, including methods for properly managing biases and preventing unintended disclosure of sensitive data and the AI systems used in the contract’s performance, including model selection, training, testing, and auditing processes.
• The contractor must include details on data sourcing and transparency, especially regarding output weighting regimes, and the responsible use of data.
• The contractor’s AI product or solution will not make or support decisions, based on the illegal consideration of race, ethnicity, gender, national origin, religion, sexual orientation, gender identity, age, nationality, medical condition, or disability in violation of Title VII of the Civil Rights Act of 1964 (Title VII), the Americans with Disabilities Act of 1990 (ADA), and the Age Discrimination in Employment Act of 1967 (ADEA), as amended.
AI Specific Conditions for LLMs
• Provide the acceptable use policy of the original developer of the AI technology.
• Provide the model, system, and/or data card of the solution that includes summaries of the training process, identified risks and mitigations, and model evaluation scores against established benchmarks.
Draft
Statement of Work Appendix – 70B03C26Q00000328
• Create and maintain end user resources such as product tutorials, developer guides, and other tools to ensure maximum utility of the AI technology.
• Develop a dedicated mechanism for end user feedback to the contractor on AI outputs that violate safety and security standards or Unbiased AI Principles.
• Describe what AI safeguards are in place for the model to respond to end user queries for which reliable information is incomplete, contradictory, or subject to individual interpretation.
• List the type of outputs restricted via content moderation and safety filters.
• Disclose any training or development of the AI technology that occurred outside the United States, to include the type of activity and the country in which it occurred.
• Disclose any modifications or configurations undertaken for the AI technology to comply with any regulation from a government other than the U.S. Federal Government.
• Describe any controls in-place to modify AI outputs applied by the vendor if the vendor is not the direct developer of the AI technology
Project Management and Documentation
The contractor shall provide:
• Project management and oversight of initiatives necessary to leverage enterprise data sets as strategic assets for the development of Artificial Intelligence applications.
• Project management and oversight of initiatives including data acquisition, defining technical data, ingest, data governance and metadata management, integration, and preparation of data, including labeling, quality assessment, when executing within DHS.
• Project status updates, risk assessments, data governance plans, and progress towards major milestones.
• Technical documentation for data sets handled and processed.
• Corresponding documentation for business process owners and end users describing metadata, handling, processing, use, and limitations of data sets.
• Documentation of compliance with NIST standards, DHS AI policy, and Unbiased AI Principles (LLMs), as appropriate.
Draft
| Global Entry Airport Modernization |
| Statement of Work (SOW) Appendix |
| 70B03C26Q00000328 |
File details come from the government source that posted it. Updated .