Bridge Policy — Interim Guidance

Effective: March 2026 | Expires: April 2026

Applies to: Staff, Contractors, and Volunteers

Foundational Rule: A person must stand behind every work product. AI is a tool, not a contributor. If you cannot explain, defend, and take ownership of a work product, it is not ready to submit, commit, or publish.

Core Principles

Principle	Description
Human Accountability	You are responsible for AI-assisted output as if you wrote every word yourself.
Accuracy First	AI outputs must be reviewed and verified before use in any work product or data pipeline.
No Unattended Pipelines	Automated AI workflows and code submissions must include a human review checkpoint before content reaches any audience or system.
Proportionate Caution	Apply review rigor appropriate to the stakes. A draft email differs from code affecting millions of records.
Data Minimization	Share only what is necessary with AI tools.
Mission Alignment	AI use must support open legal data and access to justice; avoid anything that undermines public trust in FLP's data.
When in Doubt, Ask	If you are unsure whether a particular use of AI is appropriate, ask before proceeding. Reach out to your manager or the internal AI team. The norms around AI use are evolving and no policy can anticipate every situation.

Permitted Uses

FLP Internal Staff

Use Case	Permitted?	Condition
FLP Enterprise licensed Claude	✅ Yes	Request access if needed
Other AI assistants	⚠️ Caution	Opt out of training and data collection; do not use for sensitive or private information
Drafting internal docs, emails, meeting notes	✅ Yes	Review before sending
Grant / blog / presentation drafts	✅ Yes	Verify all facts and capability claims before use
Research & summarizing research	✅ Yes	Do not disclose unreleased internal data
Sharing contract terms, partner SLAs, API pricing	❌ Prohibited	May breach confidentiality obligations
Sharing private keys or credentials	❌ Prohibited	Treat as passwords; manage through env variables
Sharing donor / funder information	❌ Prohibited
Sharing user data or PII (any source)	❌ Prohibited
Resume screening / candidate ranking	⚠️ Limited	Risk of bias; requires approved safeguards first; may be used for limited data analysis

Developers and Technical Staff

Use Case	Status	Notes
Coding assistants	✅ Yes	Read and understand all code before committing or merging
Generating boilerplate code, tests, docs, and general code completion	✅ Yes	Standard code review applies
Disclosing AI use in pull requests	✅ Required	Required for significant AI-generated contributions
Entering system architecture or private logs	⚠️ Caution	Share only what is needed; remove any PII; use FLP-licensed tools
Fully autonomous AI pipelines	❌ Prohibited	Human checkpoint required before commit/merge

Role-Specific Guidance

Developers & Technical Volunteers

Verify any AI-suggested dependency is legitimate and actively maintained before adding it.
Do not use AI to contribute to areas of any codebase you do not personally understand, except for simple self-contained changes such as typo fixes that require no knowledge of the surrounding system to verify.
Development of AI-powered tools, products, or features must adhere to the AI Development Policy.

Non-Technical Staff

Verify all capability claims, program descriptions, outcome metrics, and statistics in AI-drafted grant proposals, presentations, or blog posts against actual FLP capabilities.

Volunteers

The same prohibitions as staff apply. Process only data or code relevant to your assigned contribution area.
If using a personal AI subscription for FLP work, configure your account to opt out of training data collection where possible.

Data Handling Quick Reference

Information Type	AI Tool Use
Publicly available legal data (case law, dockets)	✅ Generally permitted
Internal process docs, wikis, guides	⚠️ Permitted with care
Staff names, roles, contact info	⚠️ Avoid — use generic placeholders
Unreleased product or data plans	⚠️ Caution until publicly announced
Donor/funder information, contract terms	❌ Prohibited
API keys, credentials, secrets	❌ Prohibited unless managed as env variables
PII of any individuals	❌ Prohibited — legal and ethical obligation

Incidents & Reporting

If AI-assisted work introduces an error into published data, code, communications, or contractual commitments, report it promptly to your manager. Include the following:

The AI tool used and its purpose
The error identified
Data or work product affected
Steps already taken

FLP will not penalize good-faith error reporting — mistakes are how we learn.

Questions? Contact: Rachel Gao — rachel@free.law

This bridge policy expires April 2026 and will be superseded by FLP's official AI Use Policy.

Artificial Intelligence Use Policy