Keystone Enters Bengaluru as AI Blurs the Lines Between Technology, Law & Competition
By Supreeth Koundinya ·
Keystone, the US-based advisory company, has established its India division, with Bengaluru as its first hub. The company plans to expand its workforce in the country over the next two years.
Founded in 2003, the company offers advisory services for legal, regulatory, tax, intellectual property and related practices in AI and technology, business and economic strategy.
Its work is supported by a network of experts across technology, economics, business, healthcare and other specialised fields, drawn from both academia and industry.
The Bengaluru team will support firms including Microsoft, Meta, Amazon, Goldman Sachs and others—both their global and Indian establishments.
Globally, the company works with Fortune 500 firms such as PepsiCo, Google, SAP, Oracle, and Meta.
One of the company’s offerings is Keystone Advanced Technology Services (K.ATS), which analyses source code and confidential big data to estimate the value of software technology for IP and tax matters.
It also conducts code reviews and forensic analyses to determine patent, copyright, trade secret, breach, and IT theft.
‘AI Brings an Enormous Ambiguity’
With AI, K.ATS also helps analyse the impact and value of specific training data on LLM performance and conducts algorithmic audits for fairness and bias concerns.
“It seemed to us a natural choice to be in Bengaluru,” says Jeff Marowits, CEO of Keystone, to AIM. “The city has enormous talent—across each of the disciplines that are required.”
Marowits said that besides software and engineering, the city also offers promising talent in management, economics, strategic advisory and consulting—particularly around competition or intellectual property relating to technology platforms.
While AI deployment is one challenge, the risks that follow have also challenged companies, especially as questions around safety are still being figured out.
This is particularly relevant when AI models develop capabilities that couldn’t have been predicted earlier in their development, and when those capabilities could complicate questions around competition and intellectual property rights.
One such problem is data valuation. If a company claims its propreitary training data gives its models a competitive advantage, that claim may be difficult for a lawyer to assess on their own. Someone needs to quantify the value of the data and its contribution to performance, and Keystone says its researchers can run experiments to measure it.
“AI brings an enormous ambiguity,” says Marowitz. “We’ve been working on machine learning for over a decade now. We have a lot of key experts in this space, and the pattern recognition of doing that in hundreds of projects.”
How Keysonte Helped Epic, IBM and Palmer Luckey Win Legal Battles
A few years ago, Epic Games sued Google over its Android app-store practices.
The allegations included requiring developers to use Google Play Billing for in-app purchases and pay commissions, making sideloading apps from outside the Play Store more difficult through Android’s warning screens, and entering agreements with phone manufacturers and developers that discouraged the use of competing app stores.
A jury unanimously found Google liable for monopolising two Android app markets in December 2023, and in October 2024 Judge James Donato ordered Google to allow competing app stores and alternative payment options on Android.
Keystone said it supported Epic throughout the case, working with its lawyers and outside economic and technical experts—from institutions such as UC Berkeley, Harvard University, and others.
Keystone helped Epic with experts across technology, economics and competition, examining Google’s Android security architecture, the technical justification for restricting alternative app stores, and the competitive effects of those restrictions.
It also helped assess whether Google’s security measures were necessary and whether less restrictive alternatives could preserve Android’s security while allowing competing app stores and distribution channels
In another example, in 2024-25, Keystone helped IBM in a legal battle against LzLabs over alleged reverse engineering of IBM’s mainframe software. IBM had alleged that LzLabs used software obtained through a 2013 licensing agreement with IBM to develop its competing ‘Software Defined Mainframe’ product.
The case went to trial in 2024, and in March 2025, the UK High Court largely ruled in IBM’s favour, finding that Winsopia, an LzLabs subsidiary, had breached its software licence and that LzLabs and its owner had unlawfully procured those breaches.
Keystone’s K.ATS team worked on the technical side of the case, conducting source-code reviews and system testing to establish how IBM’s software was reverse-engineered.
K.ATS brought together two experts with different areas of technical expertise, producing technical reports and providing evidence on the alleged reverse engineering.
In another case, Keystone helped Oculus Founder Palmer Luckey and Meta (formerly Facebook) defend against an intellectual-property lawsuit brought by the owners of Total Recall Technologies (TRT).
TRT alleged that Luckey had used the company’s VR headset design as the basis for Oculus, which Facebook acquired for $2 billion in 2014.
Keystone’s K.ATS team investigated whether TRT’s headset design was actually unique and valuable enough to support the claim. It analysed the technical characteristics of the competing headsets, their intended uses and the software and SDKs behind Oculus products.
Keystone also brought in Ravin Balakrishnan, a computer-science professor at the University of Toronto and VR expert, who conducted an extensive technical comparison.
The analysis found that many design elements were already common among VR headsets at the time, while key differences reflected the products’ different purposes: TRT’s prototype was designed for immersive 360-degree video, whereas Oculus was designed primarily for video games.
Furthermore, Keystone and Balakrishnan argued that the TRT headset’s large number of pixels would have made it difficult for contemporary consumer graphics hardware to render games at sufficiently high frame rates, potentially causing motion sickness.
In other words, simply finding similarities between the two headsets would have missed the engineering constraints behind those designs.
The case went to trial in October 2021, and a California federal jury cleared Luckey and Facebook of wrongdoing.
Can We Predict What New Laws Will Come into Place?
Marowits adds that it is also less predictable how laws and regulations will evolve in the near future, and the detrimental effects that they can have on current products and services.
Particularly in India, where AI hasn’t reached the same maturity levels as in the United States or China, it remains unclear how the government will frame regulations and laws around it.
“We don't know where the law will go, but we know that governments and courtrooms need to grapple with this changing technology, and they need to adapt to it—while still harnessing the value of the technology,” he says.
Because of this challenge, he says Keystone helps companies describe their technologies to regulators, outline safety measures, and explain the algorithms that power their products.