AI generated candidates
Technical solutions, compounds, designs, architectures, parameters, experiments, or simulations originated in part through AI output.
Specialized module within Federal Innovation Counsel
AI-Assisted Inventorship and Human Contribution Review
For federally funded companies that use AI to generate, select, optimize, simulate, or test technical solutions, the difficult issue is often not whether AI was used. It is establishing what each human actually contributed to the claimed invention and what evidence supports that conclusion.
Current legal frame
AI use does not create a separate inventorship standard.
The ordinary standards for inventorship and conception apply whether or not AI systems were used. AI cannot be named as an inventor; only natural persons may be inventors. AI involvement also does not automatically create an inventorship defect, and no conclusion should be drawn solely from the fact that prompts, outputs, or AI tools were used.
The factual record can become difficult when AI participates in exploration, design, selection, optimization, testing, or documentation. The analysis remains claim-specific, fact-intensive, and controlled by current law and official guidance.
Read the USPTO's November 2025 Revised Inventorship Guidance for AI-Assisted Inventions.
Explanatory process—not a legal test
A practical chronology for AI-assisted technical work
- Human Problem Definition
- AI-Assisted Exploration
- Human Selection or Modification
- Testing and Validation
- Claimed Invention
- Evidence Record
Text equivalent: the module may trace how a human-defined technical problem moved through AI-assisted exploration, human selection or modification, testing and validation, the claimed invention, and the records that support human contribution. This sequence is explanatory only and is not a legal test.
Best-fit situations
When the contribution record deserves focused attention
Humans shaped outputs
Researchers modified, rejected, combined, interpreted, tested, or validated AI-generated material.
Multiple contributors
Engineers, researchers, contractors, collaborators, or AI-system developers participated in different ways.
A decision point approaches
A patent filing, financing, license, diligence event, acquisition, enforcement matter, or portfolio review is approaching.
What the module examines
Claims, chronology, contributors, records, and ownership
- Claimed or potentially claimed technical subject matter
- Chronology of human and AI activity
- Problem definition and system configuration when factually relevant
- AI outputs and human selection, modification, combination, interpretation, testing, and validation
- Technical records, communications, and contributor roles
- Employee and contractor ownership records
- AI vendor or tool terms when relevant to ownership or confidentiality
- Federal award relationships
- Patent-versus-trade-secret choices
- Evidence gaps and unresolved questions
- Appropriate evidence-preservation practices
- Current official guidance and matter-specific authority
Deliverables
A contribution map and next-action record
Human Contribution Map
A claim-oriented map of asserted human technical contributions.
AI-use chronology
A source-linked chronology of relevant human and AI activity.
Contributor questions
Inventorship, ownership, contractor, and collaborator questions that remain open.
Evidence preservation
Recommendations for preserving records without implying that prompts alone decide inventorship.
Unresolved-fact list
Facts that cannot responsibly be concluded from the available record.
Protection action plan
Patent, trade-secret, confidentiality, documentation, or referral actions within scope.
Future intake workflow
Recommendations for recording AI-assisted invention activity going forward.
Coordination plan
Defined next actions for management, technical personnel, patent counsel, and specialists.
Integration
Add it to the Portfolio Review—or scope it to a specific event
The module can be added to a Federal Innovation Portfolio Review when AI materially participated in the inventive process, or it can be separately scoped for a specific filing, diligence event, or portfolio question.
See the Portfolio ReviewBoundaries
What this module does not promise
- No guarantee of correct inventorship, validity, enforceability, or patentability
- No assumption that AI use created a defect
- No formal opinion, correction, prosecution action, dispute, or transaction review without separate scope
- No definitive conclusion when missing records leave material facts unresolved
Focused discussion
Describe the decision point without sending invention details or confidential technical records.
Fit, conflicts, competence, scope, records, contributors, budget, and timing must be evaluated before substantive review.