The U.S. Government and Accountability Office has published a short and helpful paper on the basics concerning Generative AI. The paper describes the operation of Generative AI in relatively simple language. Additionally, the paper raises, at a high level, some of the pressing concerns related to Generative AI, such as bias, intentional and unintentional misinformation creation, copyrighted material used by Generative AI and privacy concerns. Notably, the paper briefly addresses national security concerns with cybersecurity, including the use of AI to "rewrite code" to make identification of sources of attacks difficult and utilization of AI to improve cyberattacks by adversaries with low technical skills—"Generative AI: The Democratization of Cyberattacks.”
Friday 30 June 2023
Wednesday 14 June 2023
European Commission is recklessly replacing established and effective FRAND valuation and licensing practices with dubious top-down rate setting
I have already made various public comments on a draft Proposal for Regulation of the European Parliament and of the Council establishing a framework for transparent licensing of standard essential patents, including the associated draft Impact Assessment report that were leaked ahead of their public launch on 27 April 2023. These comments were first published in IAM (paywalled) and then republished in IP Finance.
Among numerous legal and other issues in these proposals, I am focusing my feedback to the European Commission, in a new consultation running 27th April 2023 to 9th August 2023, on the anticipated methodologies for setting aggregate and individual SEP royalty charges by the new competence centre at an expanded EUIPO.
My analysis is summarised below and my full submission including detailed analysis and support can be downloaded here.
regulation largely ignores and seemingly abandons comparable
license valuation of patent portfolios—that predominates in licensing
negotiations and court decisions—and implicitly replaces this with the dubious top-down
that is antithetical to patent law and is repeatedly rejected by the courts
worldwide. Comparable licenses provide
fair, reasonable, non-discriminatory and effective royalty benchmarks in
bilateral and multilateral SEP valuation and license charging. The standing of
these—typically global—benchmarks is underpinned by billions of dollars of licensing
income over decades in numerous licenses including many licensors and licensees.
If a prospective
licensor can demonstrate that it has infringed and valid patents it is entitled
to a license. If these are standard-essential patents it is obliged to offer
and is entitled to receive FRAND royalties. Where charges and other terms have
been established in existing licenses, some of these can often be comparable
benchmarks for licensing other implementers.
legislation makes only one passing mention of comparable licenses when
describing difficulties including transparency and complexity in making FRAND
determinations. The impact assessment only includes references to comparable
licenses to acknowledge that they are used and to indicate that some are dissatisfied
with the extent of disclosure of existing licensing terms and licenses. Neither
document finds that the established royalty charges in existing licenses are excessive
or inapplicable FRAND licensing benchmarks.
Elsewhere in licensing
negotiations and in litigation, comparable licenses are generally considered to
provide the very best benchmarks in determining royalty charges.
Setting aggregate rates
and apportioning them among patent owners, centrally by the EUIPO—even on a
non-binding basis—will unnecessarily distort the free market processes in standards
development and FRAND patent licensing compensation that has been effective in
enabling the world’s fastest growing and largest ever technology ecosystem
serving more than five
billion people and 16 billion
connections with cellular worldwide. Parties
in licensing disputes will feel obliged in the proposed mandatory conciliation
process to give significant weight to the EUIPO’s determinations, as will the
courts. However; there is no basis whatsoever, let alone supporting evidence,
to infer there is market failure or harm to be fixed, or that established
benchmarks for royalty charges need to be replaced.
Despite the existence of
well-established licensing benchmarks, there is significant dispute about how else
to value SEP portfolios and determine royalty charges for these. According to
the impact assessment, “Although an impressive amount of scholarship has
analysed or interpreted the FRAND concept, this scholarship is characterized by
persistent differences of opinion on key aspects of the FRAND concept such as
royalty evaluation methods and obligations to license certain parts of the
 The Commission is explicit in its intention to determine aggregate royalties for some technology standards. As the impact assessment indicates from the results of its literature analysis: “An aggregate royalty for a standard is the royalty due for all SEPs on the standard. It is the starting point in a top-down determination of the royalty to be paid for a given portfolio.” The Commission’s desire that the essentiality of all patents in standard, or a representative random sample of them, are checked, rather than only small numbers of them per patent owner, stealthily implies that it wants patent counts to be used as measures of patent strength— as required in the top-down approach—even though this widely contested apportionment method is not explicitly identified or advocated in the proposed legislation. The proposed legislation requires that “The checks will be conducted based on methodology that ensures a fair and statistically valid selection capable of producing sufficiently accurate results about the percentage of truly essential patents among each SEP holder's registered SEPs.” The impact assessment also hopes that “if the register will be perceived by SEP holders as a means of indicating portfolio strength (and e.g. used in negotiations to determine the share of aggregate royalty applicable to them), they may increase the number of registered patents.”