AI & Startup Law
What is different about forming and funding a startup when the company is an AI company?
The fundamentals of forming a startup — entity choice, founder equity and vesting, the 83(b) election, IP assignment — are the same for an AI company, but two of them carry AI-specific traps. The IP assignment must explicitly capture model weights, fine-tuning artifacts, and training data, and fundraising diligence will probe data provenance and open-weight license thresholds that can reprice a round. Lysinski & Associates P.C. handles both the standard formation and the AI-specific layer.
What are the first legal steps for an AI startup?
The same core steps as any venture-track startup: choosing an entity (often a Delaware C-corporation for a company that expects to raise from VCs), issuing founder equity with vesting, making the 83(b) election on time, and getting a clean IP assignment from every founder and contractor. The firm's Startup Law resources cover these fundamentals in depth.
What is different for an AI company is not the checklist — it is the scope of the IP assignment, covered next.
Does my IP assignment cover my model weights and training data?
A standard software IP assignment often does not say so explicitly. For an AI company, the assignment should name model weights, fine-tuning artifacts, embeddings, prompts, and training data specifically, so there is no gap an investor's engineer can find later.
This is the single most common AI-formation gap. See the due-diligence and AI-generated-code pages.
Why does the 83(b) election matter, and when?
When founders receive equity subject to vesting, an 83(b) election lets them be taxed on the value at grant rather than as it vests; it generally must be filed with the IRS within 30 days of the grant, and the deadline is unforgiving.
It is the same rule for any startup, but it is easy to miss in the rush to ship an AI product — calendar it at formation.
What changes when VC or private equity comes in?
The mechanics are standard — a term sheet setting valuation, liquidation preference, and board and control terms, and a choice between a SAFE and a priced round, with the resulting dilution. What is added for an AI company is an AI-specific diligence layer on top of the usual financial and corporate review.
The structure is core startup law; the diligence overlay is where this firm's AI work applies.
What AI-specific diligence can reprice a round?
Gaps in chain of title for data and weights, open-source contamination in the codebase, undisclosed open-weight license obligations (for example, Meta's Llama Community License requires a separate license above 700 million monthly active users and 'Built with Llama' attribution), and unbounded autonomous-agent liability.
Clean these before the term-sheet clock starts. The due-diligence, open-source, and agent-liability pages go deeper on each.
Talk to an attorney who builds AI
You are about to raise, and your cap table and entity are clean — but is your IP assignment written so it actually covers your model weights and training data? Get formation and the AI-specific layer handled together, before diligence. (773) 777-9888.
For the firm’s related legal service, see AI startup & product counsel.
Frequently asked questions
What entity should an AI startup use?
A company planning to raise institutional venture capital is most often formed as a Delaware C-corporation, because that is the structure investors expect; the right answer depends on your funding plans and goals. The firm's Startup Law resources cover entity choice in detail — this page focuses on what an AI company adds on top.
Do I need to do anything special with IP assignments for an AI company?
Yes. Make sure assignments from founders and contractors explicitly name model weights, fine-tuning artifacts, embeddings, and training data, not just 'software' and 'inventions.' A generic assignment can leave the company's most valuable AI assets ambiguously owned.
What is an 83(b) election and why does it matter for founders?
It is a tax election that lets a founder be taxed on restricted equity at its grant-date value rather than as it vests. It generally must be filed within 30 days of the grant, and missing that window can create a much larger tax bill later. This is standard startup tax planning, not AI-specific — but it is easy to overlook.
SAFE or priced round for an AI startup?
It depends on stage and leverage: SAFEs are common and fast at the earliest stage, while a priced round sets a valuation and gives investors equity and governance rights now. The AI-specific diligence applies either way, so the instrument choice is a standard startup-finance question layered on top of clean AI chain of title.
What do AI investors diligence that other investors don't?
Training-data provenance, whether IP assignments cover model weights and fine-tuning, open-source and open-weight license compliance, and exposure from autonomous-agent features. These sit on top of the usual corporate and financial diligence and can move valuation or stall a round if they surface unprepared.
Should I bring in a lawyer at formation or at the raise?
At formation. Fixing entity, equity, and IP-assignment gaps early is far cheaper than discovering them during diligence with a term sheet on the clock — especially the AI-specific assignment scope, which is awkward to repair after contractors and co-founders have moved on.
More in AI & Startup Law
- AI & Startup Law — overview and all topics
- AI Startup Due Diligence
- Who Is Liable When an Autonomous AI Agent Acts? (And What Guardrails Hold Up)
- AI Product Terms of Service, Acceptable Use and Disclosure Requirements
- Open-Source vs Open-Weight AI
- Is AI-Generated Code Safe to Use Commercially? (Vibe-Coding Legal Review)
- AI Compliance for Regulated Industries
- Which AI Laws Affect My Startup? A Multi-State and EU Compliance Map (2026)
- Internal AI-Use Policy and Shadow AI
- Hiring a Lawyer for Your AI Startup