NIST's voluntary risk framework asks for the model documentation a patent application needs anyway

2023-01-26In-House

National Institute of Standards and Technology, "Artificial Intelligence Risk Management Framework (AI RMF 1.0)," NIST AI 100-1 (January 2023), released January 26, 2023, with the companion AI RMF Playbook

The National Institute of Standards and Technology released version 1.0 of its Artificial Intelligence Risk Management Framework, NIST AI 100-1, as directed by the National Artificial Intelligence Initiative Act of 2020. The document states that it "is intended to be voluntary, rights-preserving, non-sector-specific, and use-case agnostic," and it was developed with the private and public sectors. Part 1 describes seven characteristics of a trustworthy system: valid and reliable, safe, secure and resilient, accountable and transparent, explainable and interpretable, privacy-enhanced, and fair with harmful bias managed. Part 2 is the Core, "composed of four functions: GOVERN, MAP, MEASURE, and MANAGE," each divided into categories and subcategories, and the Institute states that "GOVERN is a cross-cutting function that is infused throughout AI risk management." The Map and Measure subcategories are largely documentation requirements. MAP 1.1 calls for the intended purposes and prospective settings of the system to be understood and documented; MAP 2.3 calls for scientific integrity and test, evaluation, verification and validation considerations, including data collection and selection, to be "identified and documented"; MEASURE 2.1 states that "test sets, metrics, and details about the tools used during TEVV are documented"; MEASURE 2.9 states that "the AI model is explained, validated, and documented." The framework treats intellectual property as a third-party infringement risk under GOVERN 6.1 and MAP 4.1, and notes that training data "may also be subject to copyright."

What it changesA patent application on a model needs, under Section 112 of Title 35 of the United States Code, a description of what the system does, the data it was trained on, how it was evaluated and what the results were; a trade secret program needs a record of what the company treated as confidential and the measures it took. The documentation the framework calls for under Map and Measure is that same record. The practical move for an in-house program is to build one documentation process that produces the engineering record once and routes it to three readers: the risk function, patent counsel for the specification and the inventorship record, and the trade secret program for its inventory of what is protected. Where the framework is adopted as a compliance baseline, the engineering team's documentation deadline becomes patent counsel's intake date. The framework is voluntary and creates no legal obligation; its value for the patent program is that it gets engineers to write things down.

More on Running an in-house patent program · AI regulation and policy

← All recent developments · Search all developments

Burbage Law is the practice of Jon Burbage, a registered patent attorney. 342 United States patents have issued from applications he wrote; the full list is on the Patents page. More about Jon Burbage.

This is general information about a published decision, agency guidance or piece of legislation. It is not legal advice and does not create an attorney-client relationship. Items concerning offices outside the United States are reported for information only and are not counsel on the law of those countries.