November 2023

The Debate Over the PREVAIL Act in PTAB Patent Reform

Patent,

by Dennis Crouch

On Nov 8, the US Senate Judiciary IP Subcommittee held a key hearing on the PREVAIL ACT – The “Promoting and Respecting Economically Vital American Innovation Leadership Act.” PREVAIL is a bipartisan proposal from Senators Coons (D-DE) and Tillis (R-NC) who are looking to strengthen the power of patents. The proposal here is directed toward AIA trials — generally attempting to make it more difficult to invalidate patent claims via IPR.  Key provisions include:

  • Imposing standing requirements to restrict who can file AIA patent challenges.
  • Requiring “clear and convincing” evidentiary standard for invalidating claims.
  • Broadening estoppel. Challengers must either challenge 102/103 invalidity via PTAB or court, not both (only on prior patents or printed publication grounds).
  • Limiting repeat challenges of the same patent.
  • Adding additional transparency and safeguards.

Senator Coons opened the hearing, laying out his views on the need for patent reform. He argued that the current system allows large corporations to “weaponize” the PTAB to invalidate patents held by small inventors and startups.

The witnesses were Lamar Smith, former House Judiciary Chair and namesake of the Leahy-Smith America Invents Act (now at Akin Gump); Joe Matal, former interim PTO Director (and until recently at Haynes & Boone), Joe Kiani, Founder and CEO of the medical device company Masimo, and Michelle Armond, partner at Armond Wilson.

Smith and Kiani testified strongly in favor of the PREVAIL Act while Matal and Armond provided support for IPRs largely in their current iteration, although allowing for limited changes.

Any legislation needs a good personal compelling story.  Joe Kiani provided that here.  Kiani testified that his pulse oximetry technology was copied and patents challenged once he proved that the technology worked. Masimo spent over $13 million defending patents at the PTAB that withstood pre-AIA scrutiny. Kiani argued the AIA’s intent was an efficient second look at patents, not reflexive, duplicative challenges by “opportunistic” big companies to invalidate smaller entities’ rights. Predictable property rights drive investment and innovation, and Kiani endorsed PREVAIL’s reforms for providing more certainty and reducing PTAB abuse. He concluded with a statement that action is needed, otherwise “the United States will continue to cede technological supremacy to China.”

Mr. Smith testified that the PTAB has strayed from its original purpose as an alternative to litigation. He said it has become a tool for infringers to undermine valid patents. He endorsed the PREVAIL Act’s reforms as preserving strong patent rights while still allowing for challenges to bad patents.

Smith similarly testified that IPR proceedings have strayed from their original purposes and have now become an additional tool for “litigation gamesmanship.” Smith endorsed the proposed legislation as supporting smaller companies and individuals against “well-resourced corporations.”  “[I]f we lose sight of creating a system that truly rewards investment in innovation, we are really hurting ourselves.”

Matal testified that PTAB proceedings have proven to be timely, cost-effective, and produce accurate patentability results. He cited a study finding the PTAB is affirmed on appeal much more often than district courts due to judges’ technical expertise. Matal argued PREVAIL’s restrictions on PTAB reviews are contrary to the public interest. He opposed limits on serial challenges, saying they could entrench bad examiner decisions. He argued PREVAIL’s standing requirement would prevent beneficial challenges. Matal concluded the PTAB provides reliable review American manufacturers need, saying recent cases like VLSI v. Intel show billions in damages can be awarded on invalid patents absent PTAB access.

Finally, Armond testified that PTAB proceedings have transformed patent litigation over the past decade.  She suggested that PREVAIL’s early forum choice requirement and aligned evidentiary standard could advance the goals of efficiency and certainty. However, Armond opposed limits on PTAB access, saying strong patents come from quality examination rather than less scrutiny. She suggested a better area for reforms is to ensure examiner resources and technical training.

The divided views expressed at the hearing reflect the reality that passage of the PREVAIL Act in its current form is unlikely this Congress. While Chairman Coons and Ranking Member Tillis aim for bipartisan patent litigation reform, these issues have entrenched stakeholders on both sides — making any major movement in Congress unlikely.

In re TikTok: Fifth Circuit Implicitly Backs Federal Circuit’s Use of Mandamus to Transfer Cases Out of W.D.Tex.

Patent

by Dennis Crouch

The 5th Circuit Court of Appeals recently decided an important convenient venue case.  Granting TikTok’s mandamus petition to have its trade secret case transferred from W.D.Tex to N.D.Cal.  The decision does not really break ground in the venue/mandamus debate, but does solidify the Federal Circuit’s parallel approach in patent cases out of the same district.  In re TikTok Incorporated, — F.4th —-, 2023 WL 7147263 (5th Cir. 2023).  Decision.

A Chinese company – Beijing Meishe Network Tech.Co. (“Meishe”) – sued TikTok for infringing its copyrights and misappropriating traded secrets associated with AV editing software.  Although both companies are Chinese in origin, the plaintiffs sued in Judge Albright’s Waco Texas courtroom.  Judge Albright is seen as tech-plaintiff friendly and so likely was the key driver for the filing location. Still, TikTok has numerous employees within the district, including 300 in the Waco area.  Further, most of the actions associated with the underlying claims took place in China — and so the plaintiffs argued that there was no particular US court that would be clearly better.   The reality though is that TikTok has most of its US employees in California.

Judge Albright denied TikTok’s motion to transfer venue and TikTok then petitioned the 5th Circuit for a writ of mandamus who has now granted the motion and ordered the case transferred to N.D. Cal.

28 U.S.C. § 1404(a) is the federal statutory provision that governs requests to transfer venue in civil cases.  It applies particularly in situations where venue in the original court is “proper,” but still inconvenient or may otherwise frustrate justice. It provides that a district court may transfer a civil action to another federal district court if the court determines the transfer is warranted based on the “convenience of the parties and witnesses” and “the interests of justice.”

Under § 1404(a), district court generally has broad discretion to grant or deny a transfer motion.  But, the party requesting transfer must at least show that convenience and justice factors weigh in favor of litigation proceeding in the proposed transferee district rather than the district where the action was initially filed.  Although a district court has discretion, the Fifth Circuit has held that § 1404(a) motions should be granted if the movant demonstrates that the transferee venue is “clearly more convenient.” In re Volkswagen of Am., Inc., 545 F.3d 304 (5th Cir. 2008). This process requires weighing factors like the ease of access to sources of proof, availability of witnesses, and the connection between the venue and the events underlying the dispute.  Prior Fifth Circuit decisions have emphasized that the analysis focuses on convenience, not on “any particular circumstance in isolation.” In re Radmax, Ltd., 720 F.3d 285 (5th Cir. 2013).  Mandamus is designed as a rare mechanism similar to interlocutory appeal, and only permitted based upon “clear and indisputable” abuse of discretion. In TikTok, the appellate panel explained:

Mandamus is an extraordinary remedy that should be granted only in the clearest and most compelling cases. To obtain mandamus relief, TikTok must show (1) there are no other adequate means to attain the desired relief; (2) the right to issuance of the writ is clear and indisputable; and (3) the issuance of the writ is appropriate under the circumstances.

In re TikTok Inc., 2023 WL 7147263, at *3 (5th Cir. 2023) (internal citations omitted).  In most situations, a district court error can only be appealed after final judgment. But, inconvenient venue (absent a greater due process violation) is generally not appealable post-trial and not a justification for a new trial.  What that means is that mandamus is typically the only mechanism available to obtain reversal of an erroneous inconvenient venue decision.

The Fifth Circuit granted TikTok’s petition. It held that the district court clearly abused its discretion in analyzing multiple § 1404(a) factors. Two examples: (1) The district court erred in finding that access to sources of proof was neutral when the relevant source code and TikTok employees with access were located in California, not Texas. (2) The district court abused its discretion in weighing the cost of attendance for willing witnesses and court congestion against transfer.

The In re TikTok Decision Does Not Reflect a Change in Precedent

In my view, this new decision applies the Fifth Circuit’s established § 1404(a) jurisprudence, and also solidifies the Federal Circuit’s parallel patent cases that purport to apply Fifth Circuit law.   There has been a paucity of 5th Circuit precedent on point, and so this helps back-fill that foundation.

On the one hand, the Fifth Circuit explained that mandamus should be rare. But, on the other hand, the panel still granted the mandamus and transfer.  The insight then is that rare things happen — perhaps more often than we think.  Here, the appellate court identified this case as an “unprecedented situation” with no connection between the plaintiff’s claims and the Western District of Texas.  This, the case turns on the case’s unique lack of any connection between the forum and the dispute. As the court stated, “[o]ur holding today is restricted to the outlandish facts of this case.”  The appellate panel did not mention that the plaintiffs had alleged that copyright infringement occurred in the W.D.Tex.

The Fifth Circuit’s decision in In re TikTok provides support for the Federal Circuit’s approach to venue transfer in patent cases. Like the Fifth Circuit, the Federal Circuit has shown a willingness to grant mandamus to transfer patent cases out of the Western District of Texas in situations where the convenience factors strongly favor another venue.  Although the Federal Circuit applies Fifth Circuit precedent to the issue, there has been limited Fifth Circuit authority directly on point regarding venue transfer and mandamus in civil suits.  The In re TikTok decision thus lends support to the Federal Circuit’s body of law.

Rounding the Bend: Claim Construction and the Role of Extrinsic Evidence

Patent,

by Dennis Crouch

Actelion Pharmaceuticals Ltd v. Mylan Pharmaceuticals Inc., No. 2022-1889 (Fed. Cir. Nov. 6, 2023) [decision]

The Federal Circuit’s bread-and-butter over the years has been claim constructions that often surprise or confuse district court judges. Part of the issue here is that most Federal Circuit judges have construed thousands of patents and are deeply immersed in the law of claim construction — while most district court judges see claim construction as a small part of a patent case, which itself is a small part of their overall docket. The other part of the issue, of course, is that no deference is given to the district court’s claim construction on appeal. The ensuing appeal then regularly turns out like that law school class where students seemed to provide really good answers in the Socratic game, but the professor could never resist tweaking — “almost right.”

The Actelion case involves a common situation where the patent claims include a measurement but do not state the level of precision, and the court is forced to decide whether values slightly outside the claimed range might be covered.

Actelion’s U.S. Patent Nos. 8,318,802 and 8,598,227 are directed to a new formulation of epoprostenol, a substance that naturally occurs in the human body and that is useful for treating cardiovascular diseases.  The claims appear directed to a product-by-process — a freeze dried powder (lyophilisate) formed from a solution that contains epoprostenol and that “has a pH of 13 or higher.” Mylan’s ANDA proposes a generic version that would have a pH made from a solution with pH slightly under 13.

The question for claim construction is whether the requirement of “13 or higher” pH encompasses something akin to a pH 12.6.  For abstract mathematicians, it is quite clear that 12.6 is less than 13 — the opposite of what is claimed. But, experimental scientists understand that a measurement of 12.6 might often be properly rounded to 13. The district court adopted Actelion’s proposed construction that “a pH of 13” includes values that round to 13, such as 12.5 and concluded that Mylan’s approach was infringing.  Mylan argued on appeal that the term requires an pH of “exactly” 13 — not values that are about or approximately 13.

The Federal Circuit opinion walked through various factors to help it understand the proper construction. As the court has held, the analysis begins with the claim language, as it is of primary importance. Additionally, “the specification is always highly relevant to the claim construction analysis” and is  often “the single best guide to the meaning of a disputed term.” Phillips v. AWH Corp., 415 F.3d 1303 (Fed. Cir. 2005).

Claim Language: The court examined whether the language setting a lower limit of “13” precludes rounding or not. The claim is written as a whole number, rather than in decimal form. The district court noted that this involved “two significant figures’ without implying any higher level of precision.”  At oral arguments, the patentee’s counsel explained:

[The inventor] did not say 13.0, which is a figure that would have 10 times the greater precision, and certainly didn’t say ‘exactly 13’, which is not scientifically possible in the context of pH, because you’re talking about measurement of billions of ions in a liter of water of hydrogen ions and determining their concentration.

Oral args at 18:00. On the other side, in places the specification appeared to equate 13.0 to the whole number 13.

Lack of Approximation Terms: The court noted that the claim lacks words like “about” that would imply approximation and that are often included within patent claims. As the court explained, “Unlike other claim terms, the disputed claim term lacks approximation language like ‘about.'”

Nature of pH Measurements: The court considered whether pH values can be measured precisely or inherently require some margin of error. The court noted Actelion’s argument that “it is not practically possible to measure exact pH values” because it would require counting every hydrogen ion in solution.

Specification: The specification alternately uses approximation terms like “about” or lacks decimal points when describing pH values. For example, the specification states the pH is “preferably adjusted to about 12.5-13.5, most preferably 13.” This shows the inventor knew how to use “about” to imply approximation, but did not for the “13” value.  The statement also suggests that 12.5 is different from 13.  As the court summarized, “the specification is inconsistent, also describing pH values like ‘13.0’, ‘11.58’, and ’13’ without clear precision.”

Prosecution History: During patent prosecution, the patentee had distinguished its invention from a formulation with pH 12 — arguing that pH 13 exhibited substantially more stability. But the prosecution history did not compare pH 13 with values in between 12 and 13. As the court explained, the examiner “drew a distinction between the stability of a composition with a pH of 13 and that of 12” but this did not address whether pH 13 includes values rounding to 13.

With claim construction, the Federal Circuit has ruled that courts can quite often construe claims based solely upon intrinsic evidence. But, at times courts are permitted to go further and access extrinsic evidence like scientific texts.  Here, the Federal Circuit concluded that this is a case where extrinsic evidence  is necessary to understand how a skilled artisan would view the claim language in terms of pH precision and significant figures.  As the court explained, “this case is one where the proper claim construction cannot be reached without the aid of extrinsic evidence.” The appellate panel thus vacated the claim construction and infringement judgment and remanded for the district court to consider the extrinsic evidence like scientific textbooks and perhaps expert testimony in the process of reconsidering the claim construction.

The panel consisted of Circuit Judges Reyna, Stoll, and Stark. Circuit Judge Stoll authored the opinion.

= = =

This case highlights an issue that has led to claim construction disputes in numerous cases involving measurements and precision with billions of dollars turning on so-called rounding errors. The lack of clarity stems from applicants not specifying the precision or margin of error for claim terms either in the original application or during prosecution. This ambiguity could be avoided if the USPTO were to put the onus on patent applicants to identify the precision and error margin for any measurements recited in the claims. For example, the USPTO examiner could reject the claims as ambiguous based upon the broadest reasonable interpretation and require the applicant to clarify whether “a pH of 13” allows for values that round to 13 and if so, the acceptable variance (e.g. ± 0.1, ± 0.01 pH units).

Requiring patent applicants to explicitly identify the precision and error margin for measurements during prosecution would prevent the need for courts to resort to extrinsic evidence to resolve ambiguity that should have been addressed before issuance. This would increase certainty around claim scope and reduce unnecessary litigation.

What do you think, should the USPTO consider more ways to have applicants clarify claim term precision when filing and prosecuting applications involving numerical measurements?

The Price of Principled Dissent: Fee-Shifting as a Deterrent to Legitimate Critique of Patent Doctrine

Patent

by Dennis Crouch

In re PersonalWeb Techs, — F.4th — (Fed. Cir. Nov 3, 2023)

A few years ago, I published an article along with Homayoon Rafatijo explaining our perspective of how the Federal Circuit has improperly extended the Kessler doctrine in cases like PersonalWeb Tech.  Law school classes teach two types of preclusion: res judicata and collateral estoppel. The Kessler doctrine is an additional patent-specific preclusion doctrine that the Federal Circuit applies against patentees in certain patent situation.  The basics of the doctrine is that it applies like non-mutual collateral estoppel to bar future litigation of infringement claims. But in PersonalWeb, the Federal Circuit removed some of the key safeguards of collateral estoppel — such as requiring that the issue have been actually litigated and decided  in a first case prior to precluding further litigation of the issue in a different context and against different parties. We spent a lot of time researching the history of the doctrine and its creation in the early 1900s and then walked through how the historical expansion the “claim” via notice pleading as well as expansion of non-mutual issue preclusion overtook any original justification for the doctrine.  Our article had some legs and was also awarded the Shook, Hardy & Bacon LLP Excellence in Research Award.  The Gov’t brief  in the subsequent Supreme Court petition did agreed that the Federal Circuit had improperly extended the doctrine.  Unfortunately, the Supreme Court denied certiorari in the case.

I was surprised to see today that in a 2-1 decision, the Federal Circuit thought so little of our argument (and the SG brief) that it determined PersonalWeb’s lawsuit was objectively baseless — thus serving as a basis for an exceptional case finding and attorney fee award of $5 million to the accused infringers.

I’ll begin with Judge Dyk’s dissent that sets out the problem:

The majority’s opinion rests on the remarkable proposition that PersonalWeb’s arguments were objectively baseless (and supported a fee award) despite the Solicitor General’s agreeing that those very same arguments were correct. The majority’s opinion effectively awards fees for a lack of success, which is not an appropriate use of section 285, and will likely chill legitimate advocacy.

Dyk, J. in Dissent.  The majority opinion authored by Judge Reyna and joined by Judge Lourie defended the Federal Circuit’s caselaw and appears to reject any notion that the Federal Circuit has expanded the doctrine beyond its original bounds.  Notably, the Federal Circuit did not contend with any of the arguments suggested either in our article or the SG brief.  The majority also did not explain its effective conclusion that the SG brief was itself objectively baseless.

The Federal Circuit’s newest decision PersonalWeb is concerning. By penalizing reasonable legal arguments and showing little regard for scholarly critique or even the Solicitor General’s briefing, the majority opinion signals an overt hostility towards rethinking existing patent doctrine and legitimate litigation that endeavors to challenge existing law.

Nose of Wax: The Ongoing Ambiguity in Claim Construction Standards

Patent,

by Dennis Crouch

One of the ambiguities with contemporary claim construction is how courts deal with loosely implied limitations from the specification. These do not rise to the level of ‘disclaimer’ but we’re never entirely sure whether the specification-forward approach of Phillips v. AWH will have traction.  The courts tend to strongly oppose importing limitations into the claims absent a disclaimer, but claim-language hook can often lead to interpretative narrowing.  This becomes even more interesting in situations where the patentee is using a coined term but without an express definition of scope.  In those instances it seems appropriate to read more into the the specification — but how much more?

The Federal Circuit’s decision in Malvern Panalytical Inc. v. TA Instruments-Waters LLC, No. 2022-1439 (Fed. Cir. Nov. 1, 2023) offers some guidance — and also rejects the use of prosecution history from a non-family-member application in interpreting patent claim scope.

At issue was the meaning of the term “pipette guiding mechanism” as used in a pair of patents owned by Malvern.  U.S. Patent Nos. 8,827,549 and and 8,449,175. The district court limited the term to only “manual” embodiments based on statements made during prosecution of an unrelated patent.  With that narrow construction, the accused automated devices clearly did not infringe — and the patentee stipulated as such. On appeal, the Federal Circuit vacated the district court’s claim construction, holding that the term was not limited to manual embodiments and that unrelated prosecution history should not have been used to construe the claims.

The patents relate to microcalorimeters, which are instruments used to measure the heat absorbed or released during a chemical reaction. Specifically, the patents describe an isothermal titration calorimeter (ITC) that contains several components including an automatic pipette assembly, a stirring paddle and motor, and a “pipette guiding mechanism.” The pipette guiding mechanism moves the pipette assembly between different positions, such as inserting the pipette into a sample cell for measurements or into a washing apparatus for cleaning. The key dispute was whether this pipette guiding mechanism included both manual and automatically operated embodiments. The Federal Circuit ultimately held that the plain meaning of the term encompassed both manual and automated guiding mechanisms.

District Judge Andrews (D.Del.) sided with the accused infringers in finding that the claimed “pipette guiding mechanism” should be limited only to manual operation forms.  On appeal, the TA defended the judgment and raised the following arguments: (1) The specification discloses two embodiments of the guiding mechanism, both of which are operated manually and refers to a “user skills” as an aspect of operating the machine.  (2) The specification goes on to describe other automated portions of the assembly, but did not do so with the guiding mechanism. (3) The term “pipette guiding mechanism” is a coined term without a plain meaning, so it can’t be broader than the embodiments in the specification. (4) that the patentee’s provided evidence that a product guide being distributed “included” the “relevant features” of their claimed invention.  That product guide only described manual operation.  (5) Finally, during prosecution of a separate non-family-member patent (US9103782), the patentee had argued that it was describing a ‘purely manual guiding system’ as an attempt to avoid prior art from an automated reference.   Although perhaps not a complete disclaimer, TA argued that these factors come together to narrowly construe the claim to cover only manual guiding mechanisms.  The Federal Circuit dismantled these arguments individually.

Claim Construction Analysis

The Federal Circuit began its analysis by looking to the plain and ordinary meaning of “pipette guiding mechanism” as would be understood by a person of ordinary skill in the art. The court found that the plain meaning encompassed both manual and automatic guiding mechanisms. This was based on examining the individual words in the claim term, as well as the overall context provided by the claims and specification. The court found no clear disavowal or redefinition in the specification or the arguments made during prosecution and so gave the claims their full ordinary meaning.

As to the coined-term argument, the court found that even coined terms can have a discernible plain and ordinary meaning based on the intrinsic evidence of the patent. There is some tension among the cases on this issue.  Compare Indacon, Inc. v. Facebook, Inc., 824 F.3d 1352, 1357–58 (Fed. Cir. 2016) (coined terms without a plain and ordinary meaning in the art can be limited more readily by prosecution statements) with Littelfuse, Inc. v. Mersen USA EP Corp., 29 F.4th 1376, 1381 (Fed. Cir. 2022) (“[T]he district court was correct in seeking to give meaning to the term ‘fastening stem’ by looking to the meaning of the words ‘fastening’ and ‘stem’ as used in the patent.”).  The appellate panel first noted that the doctrine is “sparingly applied” and, as here, the patent uses the individual words “pipette” “guiding” and “mechanism” in a manner that can be readily understood and does not suggest a manual limitation.  Rather, “the plain and ordinary meaning of ‘pipette guiding mechanism’ is a mechanism that guides a pipette, which can be either manual or automatic.”

In this case, the claims lack a textual hook for a manual requirement and the court found no “clear and unmistakable” disavowal.  As such, the proper construction is the full scope.

Use of Unrelated Prosecution History

A significant aspect of the Federal Circuit’s decision was rejecting the district court’s reliance on prosecution history from a non-family member patent to construe the claims. Although not an official  ‘family member,’ the parallel application had a shared assignee, was being prosecuted by the same law firm, and and also covered a pipetting system. In that non-family-member case, the patentee had distinguished its invention from the prior art by arguing that its guiding mechanism operated manually rather than automatically.  The district court treated the statements as admissions limiting “pipette guiding mechanism” to manual embodiments.  That office action rejection was then submitted to the PTO in these cases as part of an IDS.

On appeal, the Federal Circuit rejected the district court analysis, holding that  statements in an “unrelated” application do not guide meaning or serve as a disclaimer and further that submission of the OA in an IDS is insufficient to incorporate those statements for purposes of construing the claims at issue.

Although this answer here is probably correct, it makes me uncomfortable because it allows patentees to tell different stories in different cases.  Here, the cases were being handled by different examiners in different art units — it might be a different story though if the same examiner was looking through all the  similar but “unrelated” cases.

Although the logic of this case all seems clear from hindsight, I continue to be concerned that this situation offers lots of wiggle room for each side to push  for advantage — akin to a nose of wax.

Safe, Secure, and Trustworthy Artificial Intelligence

Patent, ,

by Dennis Crouch

The White House this week issued a new executive order focusing on a variety of aspects of regulating artificial intelligence, some of which focuses on IP issues.  The executive order lays out eight guiding principles for manages risks while allowing growth and benefits:

  1. Ensuring AI is safe, secure and trustworthy, including through developing guidelines, standards and best practices, verifying reliability, and managing risks related to national security, critical infrastructure and cybersecurity.
  2. Promoting innovation and competition in AI, such as through public-private partnerships, addressing intellectual property issues in ways that “Protect inventors and creators”, and ensuring market competition and opportunities for small businesses.
  3. Supporting workers affected by AI adoption, including through training, principles for workplace deployment, and analyzing labor market impacts.
  4. Advancing equity and civil rights when using AI in criminal justice, government benefits, hiring, and other areas.
  5. Protecting consumers, patients, passengers and students from risks of AI systems.  “[C]onsumer protections are more important than ever in moments of technological change.”
  6. Safeguarding privacy including through evaluating commercial data use and advancing privacy-enhancing technologies. “[T]he Federal Government will ensure that the collection, use, and retention of data is lawful, is secure, and mitigates privacy and confidentiality risks.”
  7. Improving government use of AI.
  8. Strengthening American leadership abroad to advance international cooperation on AI.

Beyond the stated goals, the order has a number of requirements — most of them directed to the various Federal executive agencies.  The tightest new regulations appear intended to focus on future AI models that are a few times larger than what OpenAI and other are currently deploying as well as “dual-use” models that could have potential national security impact.

  • It requires companies developing or intending to develop “potential dual-use foundation models” to provide information to the government about model training, ownership of model weights, results of AI “red team” testing, and measures taken to meet safety objectives.
  • It authorizes the Secretary of Commerce to define the technical conditions that would trigger these reporting requirements. Until defined, reporting is required for models trained on over 10^26 operations or primarily on biological sequence data over 10^23 operations.
  • NIST is charged with developing standards and tests to ensure that AI systems are safe, secure, and trustworthy and the Department of Commerce to develop guidance for content authentication and watermarking of AI-generated content.

The intellectual property-related aspects of the executive order ask the appropriate agencies to work on the problem.

  • It directs the USPTO Director to issue guidance to patent examiners and applicants on AI and inventorship, including issues related to using generative AI in the inventive process. It calls for the USPTO Director to issue updated guidance on patent eligibility for AI and emerging technologies.
  • It instructs the USPTO Director to consult with the Copyright Office and make recommendations on potential executive actions related to copyright and AI, including the scope of protection for AI-generated works and the use of copyrighted works for AI training.
  • It directs Homeland Security to develop a program to address AI-related intellectual property theft, including investigating incidents and enforcing actions. It also calls for updating the IP enforcement strategic plan to address AI.
  • It encourages the FTC to use its authorities to promote competition in the AI marketplace and protect consumers and workers from related harms.
  • It promotes public-private partnerships on advancing innovation, commercialization and risk-mitigation methods for AI. This includes addressing novel IP questions.

None of these requirements have immediate effect, but indicate that there will be further action over the next few months.