Here’s another ChatGPT response, this one to “What are the key provisions of the Bayh-Dole Act that federal agencies fail to enforce?” It’s sort of a trick question, and ChatGPT fails to recognize that. Given that ChatGPT must use extant literature on Bayh-Dole as its training set, ChatGPT’s response is a nice, broad depiction of what’s out there in the wild. What is impressive, then, is how closely ChatGPT’s response tracks the language used by law firms and university licensing offices in discussing Bayh-Dole compliance. In the technical sense of “having no regard for the truth,” as Harry Frankfurt has it, both ChatGPT and the university crowd, excel at “bullshit.” It is just that ChatGPT can have no regard for the truth–at least, it does not appear to be algorithmically capable of doubt, skepticism, evaluation of sources, evaluation of evidence and argument offered by those sources–but the other folks, the humans, the non-artificial intelligences should have these capabilities but apparently have largely turned them off.
Here we go. ChatGPT is in italics.
The Bayh-Dole Act is a federal law that allows organizations to retain ownership of intellectual property (IP) developed using federal funds.
The summary repeats common stuff found on-line–and chronically wrong. Bayh-Dole is specific to inventions and plant varieties, so “intellectual property” is inappropriately broad. What does it matter? Because, for one, there’s a different federal regime for copyright. And copyright is at the heart of academic publishing, and so the idea that Bayh-Dole also controls copyright runs over into the idea that Bayh-Dole also gives universities some rights in faculty and student articles and lab notebooks and the like. Bayh-Dole does not do that.
We also see a repeat of the “developed using federal funds” which is inappropriately narrow. Inventions made in work receiving federal funding, “made under contract”– where “made” has an expressly defined meaning (“conceived or first actually reduced to practice”–wording drawn from a now obsolete part of patent law dealing with establishing priority of invention, not scope of interest), as does “contract” (a funding agreement–grant, contract, or cooperative agreement–but not anything not sorted into one of these three categories). Not inventions “developed”–development comes after making, if development is necessary. “Development,” too, is a defined term in federal procurement. And the use of federal funds to develop an invention after having made it is irrelevant to a determination of whether Bayh-Dole applies to the invention itself. Just doesn’t matter. What matters is the scope of the work for which a contractor has petitioned or bid for federal support for at least some of that work. If federal funds support at least a portion of that work, then any invention made in that work, whether that is the federal portion of the work or not, becomes, when it is acquired by the contractor, a subject invention under Bayh-Dole. What matters is the work as a whole, not necessarily just the portion of the work that federal funds have been used to pay for. Work, some supported by federal funds? Invention made in scope of work? Contractor acquires? Bingo.
“Developed using federal funds” is just so wrong–and yet you will find this usage all over the internet, which should give you pause for the competence of the folks writing this way. “Developed using federal funds” is not just a loose, easier to understand shorthand–it’s just stupidly, stubbornly wrong.
“Any invention conceived or reduced to practice with the assistance of the Federal funding is subject to the Bayh-Dole Act.” Law firm “information” on compliance
“to determine ownership rights of an invention developed using federal funds” Law firm discussing Stanford v Roche
“to acquire ownership of inventions developed with such funding” North Dakota State University (Bayh-Dole doesn’t “allow” acquisition, it allows retention following acquisition–huge difference, really)
“are permitted to elect title to federally funded inventions” University of Virginia (research or development is federally funded, not inventions; “elect” title is wrong–elect to retain title that they got in some non-Bayh-Dole way)
“The University is entitled to retain ownership of any inventions created as a result of federal funding” Drexel (and wrong about “entitled”–nonsense)
“made under the auspices of federal funding” Rice (“made under contract” would work, as would “made in performance of work under contract,” but not “auspices”–goofy)
Some get this bit right, for instance:
“resulting from federally-funded research” Michigan (but they are wrong about “allows . . . ownership rights”)
“inventions made under federally sponsored research” Delaware (though “federally sponsored funding agreement”) would be better, and even better would be pointing out that “made” has a special not so obvious meaning).
We continue with ChatGPT:
This law applies to organizations that receive funding from federal agencies, such as universities and businesses.
Bayh-Dole applies principally to federal agencies, not to contractors. The one provision directed at contractors is 35 USC 202(a). That’s it. The rest is focused on what federal agencies must do in contracting and licensing. It is by federal contract–compliance with Bayh-Dole by federal agencies–that most of Bayh-Dole applies to universities. By breaching a patent rights clause, a contractor does not violate Bayh-Dole, the statute. In fact, given that 35 USC 202(a) simply offers contractors a right, there is nothing in Bayh-Dole that a contractor can violate. All contractor problems arise in the patent rights clauses.
And of course, Bayh-Dole addresses only nonprofits and small businesses–not all businesses. Executive branch patent policy–EO 12591, incorporating Reagan’s 1983 Memorandum–authorizes Bayh-Dole-like provisions for large companies. And the “funding” must come in the form of a “funding agreement,” which has an express definition in Bayh-Dole–in particular, the funding has to be for research or development. If it’s for education, procurement, investment, or anything else, it falls outside Bayh-Dole’s scope. Federal agencies use this workaround all the time, with “other transactions” (don’t call them contracts), and with contracts that specify a purpose other than research or development, even if the primary activity will be research or development.
There are several key provisions of the Bayh-Dole Act that federal agencies may fail to enforce, including:
This part is rich–federal agencies don’t enforce much of anything other than proper format of invention disclosures–they don’t even review disclosures for documentation to back determination of proper identification of funding scope.
1. The requirement that organizations disclose any IP they have developed using federal funds within a certain period of time after the development of the IP.
This bit misstates the disclosure requirement. Not “any IP” but rather subject inventions, only. You know now to see “developed using federal funds” is flat wrong. And “after the development of the IP” is just gibberish; try “within a reasonable time after it becomes known to contractor personnel responsible for the administration of patent matters” 35 USC 202(c)(1). The Bayh-Dole disclosure requirement kicks in when a contractor’s patent personnel have received a conforming disclosure from the inventor and the invention has been acquired by the contractor–contractor acquisition of an invention is necessary for an invention to be a subject invention under Bayh-Dole. Bayh-Dole originally required contractors to disclose subject inventions (ones they had acquired, made under contract) within a reasonable time. That would have required contractors–universities especially–to actively monitor inventive activity. It would be on the tech transfer office to go out and monitor federal grant work for inventions and timely report them. But in 1983, two years after Bayh-Dole went into effect, university-affiliated patent officers got the law changed so that they didn’t have to do anything until an inventor properly disclosed an invention to them. Cozy, but slack.
2. The requirement that organizations make a reasonable effort to license any IP they have developed using federal funds to potential users, in order to promote the use and dissemination of the technology.
This requirement does not exist. It is fantasy. There is no “reasonable effort” bit in Bayh-Dole or in the patent rights clauses. There is also no “incentive” stated there to license, and it would be dumb to put in either a requirement or an incentive to license because small companies–a key focus of the law–don’t have any obvious reason to license the inventions they acquire. They can just practice them, to grow their companies (or sink into bankruptcy, inventions as they often are). There is also no requirement to try to “commercialize” inventions. That’s also fantasy. Bayh-Dole’s primary focus throughout is “utilization”–and in particular, “practical application,” which is utilization with benefits available to the public on reasonable terms” (that is, terms that would be expected if there were competition, even if patents are used to suppress competition). Where “commercialization” is used in Bayh-Dole–and the law does anticipate commercialization as one way to achieve utilization, collaboration with industry, and US manufacturing goals–there’s also “public availability.” That is, a contractor can either use an invention or license it. Doesn’t much matter. No requirement to do one or the other, no incentive to do one or the other.
The “reasonable effort” fiction comes from attempts to prevent federal agencies from protecting the public from lack of availability, nonuse, and unreasonable terms by using the compulsory licensing authorization of 35 USC 203. The “march-in” provisions authorize federal agencies to request that contractors grant licenses if they have not timely achieved practical application of a given subject invention. Here’s the key bit (203(a)(1):
has not taken, or is not expected to take within a reasonable time, effective steps to achieve practical application of the subject invention in such field of use
The “is not expected to take” part might be rebutted by a showing of effort–perhaps even “reasonable effort”–but the standard is an expectation to take “effective steps” in a reasonable time. Nothing here about reasonable effort. The reasonable effort idea is just made up nonsense serving as political spin to try to shut down the public interest part of the Bayh-Dole bargain. Might say that nonsense is, well, corrupt.
Bayh-Dole does not give contractors a pass if they “try.” As Yoda has it, “Do or do not. There is no try.” Come to think of it, Yoda probably could do better at restating Bayh-Dole than either ChatGPT or a gibber of university licensing administrators.
3. The requirement that organizations share any royalties or other income they receive from licensing their IP with the federal agency that provided the funding for the development of the technology.
The royalty sharing requirement is limited to nonprofits. See 35 USC 202(c)(7)(B). Bayh-Dole distinguishes royalties from “income earned . . . with respect to subject inventions.” Royalties must be shared, but not other income earned. But all of this is beside the point! The royalty sharing is with inventors. There is no royalty sharing with the federal agency. This is just totally wrong. Drafts of other bills–Thornton, Schmitt–anticipated royalty sharing with federal agencies (a hugely bad idea, just like allowing federal agencies to have a money interest–a secret money interest at that–was and is a hugely bad idea). But Bayh-Dole does not do that. Here, ChatGPT is totally wrong. And of course “development of the technology” is off-base. Not “development,” not generally “technology,” not “funding for.” All gibberish.
4. The requirement that organizations provide the federal government with a royalty-free license to use any IP they have developed using federal funds for government purposes.
Okay. Bayh-Dole does require federal agencies to require a non-exclusive, royalty-free license. But the scope of that license is “to practice and have practiced.” The wording here comes from the Kennedy patent policy and the Institutional Patent Agreement master contract. There, “practice” is defined as “to make, to use, and to sell”–that is, the substantial rights one may have in an invention. The license, then, is not merely “to use” an invention, but to practice it–all the rights are available to the government but for the right to exclusively license or assign the entire interest in an invention or to enforce the patent. The restriction “for government purposes” does not show up in Bayh-Dole, though “Governmental purpose” is used in the Kennedy patent policy. The Bayh-Dole license to the government is general, broad, unlimited. Anything the federal government can do, that’s the scope of the license the contractor grants.
In general, federal agencies are responsible for enforcing these provisions of the Bayh-Dole Act . . .
Nope. Federal agencies are responsible for the management of patent rights clauses in individual funding agreements. The Department of Commerce is responsible for the administration of the implementing regulations and patent rights clauses (and making a slobbering mess of that). But there is no enforcement specified for Bayh-Dole. No oversight. In the most positive spin, self-regulating. What could go wrong with that?
. . . to ensure that organizations are complying with the law
Nope. Here, it should read “to ensure that contractors are complying with their patent rights clauses.” But even so, at crucial points Bayh-Dole gives federal agencies the power to ignore requirements (such as contractor reporting on utilization), waive them (such as US manufacturing), or walk away even after determining non-compliance (as for anything that involves the march-in procedure–see 37 CFR 401.6(h)). As the 1995 GAO report found, there’s no oversight. Federal agencies leave it up to universities to comply, and universities are hopelessly inapt and inept at compliance. As one director of a licensing office told me, point blank, in response to a question regarding Bayh-Dole, “We will do that if and when a federal agency makes us do that.”
. . . and that the government is able to benefit from the IP developed using its funds.
This part looks like a well constructed English clause, but it makes no sense it context. We are talking about enforcement, and the sentence starts out that federal agencies are responsible “enforcing these provisions to ensure that” [contractors are complying] and [the government is able to benefit]. It just doesn’t work. If contractors comply, then they grant the government its license. The government then benefits however federal agencies choose to exploit the government’s right under that license. Bayh-Dole has nothing to say, expressly, about how federal agencies go about practicing (including having practiced) any given subject invention. There is nothing to enforce. And federal agencies have a tough time enforcing on themselves what they don’t care to do. It just doesn’t work. Here, looks good, is gibberish. Sadly, it’s not just what one would expect from an AI text generator. It’s just like what one expects from university licensing offices and the law firms selling services to them and to companies dealing with them.
However, it is possible that some federal agencies may not always adequately enforce these provisions, potentially leading to organizations not complying with the requirements of the law.
Well, yes! Spot on. Perhaps “it is possible” is a bit timid, but let’s go with that.
As for the list ChatGPT generated, it’s bland to the point of useless. The problems of enforcement extend from failure to accurately cite federal funding in patent applications to misallocation of royalty income to failure by agencies to determine exceptional (not extraordinary or rare) circumstances and a blind eye to nonprofit assignment of subject inventions and march-in for nonuse (lots of nonuse), unreasonable (non-competitive) pricing, and lack of availability (including availability in alternative forms and combinations). There is nothing in the ChatGPT write up that would even spark a discussion regarding whether Bayh-Dole lacks necessary oversight, auditing, and enforcement, and where agency or contractor noncompliance might be harming the public. Just bland, often wrong, sentences that look great but are, well, off, sometimes way off, way wrong.