AI Assistant for City Staff Productivity and Research
What it is for
Inform
- Staff used Copilot to assist with information-intensive tasks including legal analysis of proposed legislation, reviewing stakeholder feedback on policy documents, and analyzing survey results — supporting research and knowledge work within OTI.
Planning & Decision-making
- Some outputs supported policy analysis and development, including reviewing stakeholder feedback for policy documents and legal analysis of proposed legislation, which may inform city government decisions.
- Run by
- Office of Technology and Innovation (OTI)
- Where
- No fixed location
- Kept
- Retained Not specified in the register; governed by Microsoft enterprise agreement and OTI data governance policies
- Shared with
- Accountable organization, Vendor
What it is for
From June to August 2025, select staff at New York City's Office of Technology and Innovation tested Microsoft Copilot — an AI writing and analysis assistant — to evaluate its potential value for city government work. Staff used the tool for tasks such as legal analysis, reviewing policy feedback, analyzing survey data, and drafting social media content. While most test cases were internal, some outputs may have appeared in public communications. The tool was provided by Microsoft on a trial basis and processes both the text users input and documents they attach.
What it collects and what happens to it
Data taken in
About behaviour
Identifiable data
- Users input text prompts and attached files as part of their routine workflows. The content of these inputs reflects user work activity and professional tasks within OTI.
Operational data
Anonymized data
- Files attached by users included operational materials such as proposed legislation, stakeholder feedback documents, survey data, and social media content drafts — organizational data used in the course of city government work.
Processing
Language Models
- Microsoft Copilot is built on large language models (LLMs) that generate text, summarize documents, and respond to natural-language prompts. The Copilot for Office 365 variant is additionally grounded in the OTI organizational knowledge graph.
What it does
Creating (Generative AI)
Human decides
- Copilot generates text, images, and documents in response to user prompts. Staff decide whether to use, edit, or discard the generated outputs; the AI is advisory and all consequential use is human-directed.
Understanding (Semantic AI)
Human decides
- Copilot for Office 365 uses a 'knowledge graph' of the OTI tenant to understand the structure and content of organizational files, enabling it to retrieve and summarize relevant documents and data in response to user queries.
Outputs
Generated content
Anonymized data
- The tool produced text, images, and documents in response to user prompts. Some outputs may have been used in public communications, such as social media content developed by OTI staff.
Run by
Office of Technology and Innovation (OTI)
- The Office of Technology and Innovation (OTI) deployed the Microsoft Copilot pilot program for select staff in its Research & Collaboration department to evaluate the tool's business value and management requirements.
Built by
Microsoft
- Microsoft provided 50 trial licenses for Microsoft Copilot for Office 365 and made Copilot Chat available to 40 users at no cost. Microsoft holds proprietary training data and the underlying model.
Kept for
Retained Not specified in the register; governed by Microsoft enterprise agreement and OTI data governance policies
- The register does not specify a retention period for inputs or outputs. Retention of data processed by Microsoft Copilot is subject to Microsoft's enterprise terms and OTI's internal data governance. The pilot ran from June to August 2025.
- Duration: Not specified in the register; governed by Microsoft enterprise agreement and OTI data governance policies
Shared with
Available to the accountable organization
- Outputs generated by Copilot (text, images, documents) are available to the OTI staff who produced them and may be used in internal or public-facing work products.
Available to vendor
- Microsoft, as the vendor, processes input data and prompts through its cloud infrastructure to generate outputs. The extent to which Microsoft retains or uses input/output data for model improvement is governed by its enterprise agreements and is not fully specified in the register.
Stored
Stored on 3rd Party Cloud
- Microsoft Copilot is a cloud-based service; input data, prompts, and outputs are processed and may be stored on Microsoft's cloud infrastructure (Azure). The OTI tenant configuration governs data residency within Microsoft's infrastructure.
- Duration: Not specified; governed by Microsoft enterprise agreement
How to read the colours
Can it identify you?
- Anonymized data
- Data about people with the link to who is broken. Stripped of identifiers, blurred, aggregated, or noised so this system can’t reasonably tie a record back to an individual.
- Pseudonymous data
- Each person’s data is tied to a token (hash, ID, template) that lets this system recognise the same person across events, but the token itself doesn’t reveal a name. Reidentification is possible with extra information.
- Identifiable data
- The data either contains a direct identifier (name, address, account name, recognisable face or voice, plate number) or carries a token this system uses to look up legal identity during processing.
Who completes the loop?
- Human decides
- This mode suggests; a person decides what to do next. The AI is always advisory — a human is in the loop on every decision. Example: a triage tool ranks cases for a clinician who chooses which to see first.
- Human executes
- This mode decides; a person carries out the result. Example: an optimizer plans the day’s trash-collection routes, and drivers run them.
- Autonomous
- This mode decides and acts on its own. No person reviews each decision or carries out the resulting action.
Definitions from the DTPR standard. Amber is about your data, violet about who decides. The fuller the shape and the deeper the colour, the more identifying the data or the less a person is involved.
- AI registerNYC AI Use Cases — Microsoft Copilot Pilot ProgramNYC Office of Technology and Innovation, AI Use Case Register, 2025. Agency: Office of Technology and Innovation, Department: Research & Collaboration.
- AI registerNYC AI Use Cases — Microsoft Copilot Pilot Program
- AI registerNYC AI Use Cases — Microsoft Copilot Pilot Program
- AI registerNYC AI Use Cases — Microsoft Copilot Pilot Program
- AI registerNYC AI Use Cases — Microsoft Copilot Pilot Program
- AI registerNYC AI Use Cases — Microsoft Copilot Pilot Program
- AI registerNYC AI Use Cases — Microsoft Copilot Pilot Program
- AI registerNYC AI Use Cases — Microsoft Copilot Pilot Program
- AI registerNYC AI Use Cases — Microsoft Copilot Pilot Program
- AI registerNYC AI Use Cases — Microsoft Copilot Pilot Program
- AI registerNYC AI Use Cases — Microsoft Copilot Pilot Program
- AI registerNYC AI Use Cases — Microsoft Copilot Pilot Program
- AI registerNYC AI Use Cases — Microsoft Copilot Pilot Program
- AI registerNYC AI Use Cases — Microsoft Copilot Pilot Program
- AI registerNYC AI Use Cases — Microsoft Copilot Pilot Program
- AI registerNYC AI Use Cases — Microsoft Copilot Pilot Program
- AI registerNYC AI Use Cases — Microsoft Copilot Pilot Program
- AI registerNYC AI Use Cases — Microsoft Copilot Pilot Program
- AI registerNYC AI Use Cases — Microsoft Copilot Pilot Program
- Register entryPublished by the NYC Office of Technology and Innovation. Reference b577c046. This disclosure was drafted with AI assistance.Schema: ai@2026-05-06-beta
What you can do
Ask about this system
Questions go to the NYC Office of Technology and Innovation, not the vendor.
Your rights
- Right to Algorithmic TransparencyCity staff and members of the public may seek information about how Microsoft Copilot works through OTI's public disclosures and the NYC AI use case register. Microsoft's product documentation provides general information about Copilot's underlying model and processing logic. Training data details are proprietary to Microsoft and not fully disclosed.
- Right to Be Informed of AI UseStaff participating in the pilot were informed they were using an AI system. Members of the public who received communications that may have been assisted by Copilot were not necessarily notified of AI involvement. The NYC AI use case register provides public notice of this system's deployment.
Risks and safeguards
- Reputational harmAI-generated content (including public communications such as social media posts) may contain inaccuracies, hallucinations, or misleading statements that could harm public trust or misrepresent city policy.Safeguard: The pilot was explicitly experimental; staff were encouraged to evaluate outputs critically. The register notes that most test cases did not have public impact and the tool was tested under supervision to determine management and security requirements before wider deployment.
- Loss of autonomyStaff may over-rely on AI-generated outputs for tasks like legal analysis or policy review, reducing critical human judgment.Safeguard: The pilot's informative autonomy level means staff used Copilot as an assistive tool; outputs were not automatically acted upon. The pilot design required staff to test and evaluate outputs as part of assessing the tool's value.