ZeMing M. Gao

👨‍⚖️💡 Uniswap Labs Responds to SEC’s Wells Notice with a Powerful Rebuttal 🔍 Uniswap Labs has responded to the SEC’s Wells notice, calling the case “weak and wrong.” Here are their key arguments: 1️⃣ The Interface Is One of Many Applications that Allows Users to Access the Protocol. - Corresponds to my explanation of the frontend and protocol being distinct and that trades can be executed through other interfaces. 2️⃣ The Protocol Is an Automated Market Maker Technology Controlled by No Individual or Entity. - Matches my point about the protocol being open-source and not controlled by Uniswap Labs. 3️⃣ Under the Plain Language of the Exchange Definition, Labs Does Not Operate an Exchange. - Reflects my point about labeling Uniswap as an "exchange" challenging longstanding SEC guidance. 4️⃣The UNI Token Is a Governance Token that Allows Holders to Control the Limited Modifiable Aspects of the Protocol. - Similar to my analysis using the Howey test, highlighting the governance role of the UNI token. 5️⃣Labs Does Not Operate a Securities Exchange as Defined by the Text of the Exchange Act. - Matches my assertion that the SEC’s claim about unregistered broker activities does not make sense. 6️⃣Recent Precedent Establishes that Labs Does Not Meet the Definition of a Broker Under the Exchange Act. - Aligns with my reference to the Coinbase vs. SEC case where a self-custody crypto wallet was not deemed a broker. 7️⃣Labs Does Not Act as a Depository Because it Does Not Take Custody of Users’ Tokens. - Matches my mention of trades occurring on the Ethereum network outside of Uniswap Labs' control. 8️⃣Labs Does Not Act as an Intermediary that “Moves” Assets Because it Does Not Touch Users’ Tokens. - Similar to the point above about Uniswap Labs not controlling the actual settlement of trades. ➕Additional Points: - Autorouter and the non-custodial nature of Uniswap Labs. - The Autorouter Is an Open-Source Tool that Recommends the Best Trading Path on the Protocol. 🔍 My previous analysis was correct and closely aligned with the key points raised in Uniswap Labs’ Wells submission. My insights on the separation of the protocol and frontend, the role of the UNI token, and the impact of the SEC’s actions on innovation and legal precedents are consistent with Uniswap Labs' response. 📜 For more details: - [Uniswap Labs Wells Notice Response on the 43 pages] https://lnkd.in/dTy992uz - [My Previous Analysis in 20 minutes] https://lnkd.in/daydy6k2 #Crypto #Blockchain #Innovation

22

2 Comments

I remain skeptical on the value of analytics and AI to make a difference to case selection and loss rates generally (w some exceptions). Essentially, there is a data completeness issue and far too much idiosyncratic case risk and too many human variables that are simply impossible to predict. I have always felt that if there was one industry that could extract value from data it would be the insurance industry which has access to the most amount of private & confidential case resolution data (as an industry) as they are often the counterparty in many cases because of their position as insurer. However, I do wholeheartedly agree with Christopher Bogart that anything that can be done by litigation funders to decrease their loss rate will make a meaningful difference to a fund's overall results (8% is impressive!). From the thousands of data points on case outcomes that I have seen (admittedly skewed because they are only resolved cases - withdrawals, losses, awards (via resolution & judicial decision)) the industry has about a 30% loss rate (I would say the loss range is 25-35% (as measured by # of cases which closely correlates with committed capital) and that cuts across all jurisdictions where there is a meaningful number of case outcomes). This means of course that the remaining 70% of the capital (probably close to 50% when you remove the committed capital used for manager fees and expenses and factoring in a <100% deployment rate) not only has to produce a return for its own commitments, it also has to overcome the actual losses and produce a return for that portion of the portfolio and the portion used to fund expenses. As Chris correctly points out, most losses are complete losses (although not always) and this puts a lot of pressure on the fund and its manager to overcome those losses. In some instances where those case losses happen early on in the life of the portfolio it could put pressure on the manager to try a 'hail mary' to try and create a higher than average outcome to overcome the loss, but in doing so it likely means they are assuming additional risk and if that decision goes badly they may not live to see another day. So, it is in investors' best interest to ensure the manager is doing everything it can to mitigate the losses, conserve cashflow at the fund level (watch the expenses and the fixed overheads), manage duration risk and actively risk share (with insurers or others). On the loss mitigation front, while AI & Data Analytics may have a role to play I believe it ultimately suffers from a data issue (topic for another post but essentially the problem is completeness of resolved case data, which is the majority of data and remains confidential) and so managers should focus on those items within their control like portfolio construction / diversification, withdrawal conditions in their funding contracts, minimum return requirements on their committed equity, case budget management and adherence, risk sharing, etc.

10

3 Comments

The legal crypto community in the US should be studying the recent ruling from the N.D. CA judge holding that LIDO DAO is a partnership and can be sued as such. Yes, it is a blow to the decentralized finance, but on the other hand, is it fair for the DAOs to exist outside of the existing laws and assume no liability for their actions? There is a lot more to study and think about, but in the meantime, I just wanted to share the news with you. https://lnkd.in/ejZSAw8m

19

1 Comment

🚀 Today's AI tool created by a lawyer is a VC investment document reviewer. It reviews the VC document given to it per a customized playbook—identifying non-market terms like participating liquidation preferences, cumulative dividends, and redemption rights. 📄✅ 🌟 Try it for free and discover how easy it is to create an automated playbook yourself. No need to hire engineers, spend a big budget, or wait for months. You can do it yourself for just $1 and 1 hour of your time. Since you already know how to do the job, you just need to write your expertie down as instructions to the AI. No coding required. 🤖💼 ✨ There is no better way to learn AI than by using it. It’s also a lot of fun to turn your legal expertise into an AI tool that automates your task. After all, as the owner of both the expertise and the use case, you are the best person to use AI to innovate. 🔧🎓

3

#AILawsuits are starting to pile up. Any one of these could be the first domino to tip which is why it is so important to build your tech stack model agnostic - you should have the ability to use any model you want without being beholden to one. #ownyourai #codeandcounsel #legaltech #ai #copyright #opensource

3

🇺🇸 The enforcement approach continues in full swing. The CFTC has once again taken action in the DeFi space by issuing an order against Uniswap. Is it even surprising at this point? Have we reached a stage where enforcement has become more predictable than any genuine regulatory efforts to create a clear and effective framework? 📌 “𝘜𝘴𝘪𝘯𝘨 𝘰𝘶𝘳 𝘦𝘯𝘧𝘰𝘳𝘤𝘦𝘮𝘦𝘯𝘵 𝘢𝘶𝘵𝘩𝘰𝘳𝘪𝘵𝘪𝘦𝘴 𝘢𝘨𝘢𝘪𝘯𝘴𝘵 𝘋𝘦𝘍𝘪 𝘱𝘳𝘰𝘵𝘰𝘤𝘰𝘭𝘴 𝘪𝘯𝘴𝘵𝘦𝘢𝘥 𝘰𝘧 𝘱𝘳𝘰𝘷𝘪𝘥𝘪𝘯𝘨 𝘤𝘭𝘢𝘳𝘪𝘵𝘺 𝘵𝘩𝘳𝘰𝘶𝘨𝘩 𝘯𝘰𝘵𝘪𝘤𝘦-𝘢𝘯𝘥-𝘤𝘰𝘮𝘮𝘦𝘯𝘵 𝘳𝘶𝘭𝘦𝘮𝘢𝘬𝘪𝘯𝘨 𝘳𝘪𝘴𝘬𝘴 𝘥𝘳𝘪𝘷𝘪𝘯𝘨 𝘳𝘦𝘴𝘱𝘰𝘯𝘴𝘪𝘣𝘭𝘦 𝘋𝘦𝘍𝘪 𝘥𝘦𝘷𝘦𝘭𝘰𝘱𝘦𝘳𝘴 𝘰𝘷𝘦𝘳𝘴𝘦𝘢𝘴 𝘵𝘰 𝘤𝘳𝘦𝘢𝘵𝘦 𝘣𝘶𝘴𝘪𝘯𝘦𝘴𝘴𝘦𝘴, 𝘫𝘰𝘣𝘴, 𝘢𝘯𝘥 𝘦𝘤𝘰𝘯𝘰𝘮𝘪𝘤 𝘢𝘤𝘵𝘪𝘷𝘪𝘵𝘺 𝘢𝘸𝘢𝘺 𝘧𝘳𝘰𝘮 𝘵𝘩𝘦 𝘜𝘯𝘪𝘵𝘦𝘥 𝘚𝘵𝘢𝘵𝘦𝘴; 𝘭𝘦𝘢𝘷𝘪𝘯𝘨 𝘣𝘦𝘩𝘪𝘯𝘥 𝘵𝘩𝘦 𝘣𝘢𝘥 𝘢𝘤𝘵𝘰𝘳𝘴 𝘢𝘯𝘥 𝘤𝘳𝘪𝘮𝘪𝘯𝘢𝘭𝘴 𝘸𝘩𝘰 𝘢𝘳𝘦 𝘰𝘯𝘭𝘺 𝘪𝘯𝘵𝘦𝘳𝘦𝘴𝘵𝘦𝘥 𝘪𝘯 𝘵𝘢𝘬𝘪𝘯𝘨 𝘢𝘥𝘷𝘢𝘯𝘵𝘢𝘨𝘦 𝘰𝘧 𝘈𝘮𝘦𝘳𝘪𝘤𝘢𝘯 𝘤𝘪𝘵𝘪𝘻𝘦𝘯𝘴” This is spot on, and we've seen a similar pattern with securities laws applying to certain digital assets, where the SEC has provided little clarity. Defining legal boundaries through enforcement actions leaves companies guessing about the classification of their products and assets, along with the regulatory requirements they must follow. This uncertainty is not a sustainable approach. 📌“𝘐 𝘴𝘩𝘢𝘳𝘦 𝘵𝘩𝘦 𝘊𝘩𝘢𝘪𝘳𝘮𝘢𝘯’𝘴 𝘤𝘰𝘯𝘤𝘦𝘳𝘯 𝘵𝘩𝘢𝘵 “𝘵𝘩𝘦 𝘤𝘶𝘳𝘳𝘦𝘯𝘵 𝘵𝘳𝘢𝘫𝘦𝘤𝘵𝘰𝘳𝘺 𝘪𝘴 𝘶𝘯𝘴𝘶𝘴𝘵𝘢𝘪𝘯𝘢𝘣𝘭𝘦,” 𝘢𝘯𝘥 𝘵𝘩𝘪𝘴 𝘤𝘰𝘯𝘤𝘦𝘳𝘯 𝘪𝘴 𝘩𝘦𝘪𝘨𝘩𝘵𝘦𝘯𝘦𝘥 𝘸𝘩𝘦𝘯 𝘸𝘦 𝘶𝘴𝘦 𝘰𝘶𝘳 𝘦𝘯𝘧𝘰𝘳𝘤𝘦𝘮𝘦𝘯𝘵 𝘳𝘦𝘴𝘰𝘶𝘳𝘤𝘦𝘴 𝘵𝘰 𝘪𝘯𝘷𝘦𝘴𝘵𝘪𝘨𝘢𝘵𝘦 𝘋𝘦𝘍𝘪 𝘢𝘤𝘵𝘪𝘷𝘪𝘵𝘺 𝘸𝘩𝘦𝘳𝘦 𝘵𝘩𝘦𝘳𝘦 𝘪𝘴 𝘯𝘰 𝘥𝘰𝘤𝘶𝘮𝘦𝘯𝘵𝘦𝘥 𝘩𝘢𝘳𝘮 𝘰𝘳 𝘯𝘰 𝘴𝘪𝘨𝘯𝘪𝘧𝘪𝘤𝘢𝘯𝘵 𝘱𝘳𝘰𝘧𝘪𝘵 𝘵𝘰 𝘵𝘩𝘦 𝘋𝘦𝘍𝘪 𝘱𝘳𝘰𝘵𝘰𝘤𝘰𝘭 𝘴𝘵𝘦𝘮𝘮𝘪𝘯𝘨 𝘧𝘳𝘰𝘮 𝘵𝘩𝘦 𝘢𝘤𝘵𝘪𝘷𝘪𝘵𝘺 𝘢𝘵 𝘪𝘴𝘴𝘶𝘦, 𝘢𝘴 𝘸𝘦𝘭𝘭 𝘢𝘴 𝘸𝘩𝘦𝘳𝘦 𝘸𝘦 𝘴𝘦𝘦 𝘱𝘳𝘰𝘢𝘤𝘵𝘪𝘷𝘦 𝘴𝘵𝘦𝘱𝘴 𝘵𝘰𝘸𝘢𝘳𝘥 𝘤𝘰𝘮𝘱𝘭𝘪𝘢𝘯𝘤𝘦.” Pig Butchering scams have reportedly resulted in $75 billion in losses. Additionally, new FTC data this week highlighted a sharp increase in consumer losses from Bitcoin ATM scams. Yesterday, the FBI issued an alert that North Korea is conducting advanced social engineering attacks on DeFi and cryptocurrency companies to steal digital assets.  It’s clear that resources and taxpayer money could be better spent addressing these significant threats, rather than enforcing for the sake of enforcement. What's the next chapter in the regulation-by-enforcement saga? 🔗 https://lnkd.in/dTvQJijD

39

2 Comments

🚀 This paper is a game-changer for anyone passionate about legal tech and AI! (Might be boring for other people 😂 ). 📚 It explores how visual cues in documents can dramatically enhance the parsing of legal texts. 🧠 By leveraging visual structure, you can quickly extract key information, making sense of legal hierarchies with ease. 💼 Best of all, this resource is free and open source, making it an invaluable tool for creating the content you need! 🔓✨ #LegalTech #AI #OpenSource https://lnkd.in/etjKpKYt Leonard Park Harvey Charles Uniman Ofer Mendelevitch Spellbook The LegalTech Fund Kitty Boxall

35

4 Comments

An (ironic) experiment with NotebookLM to learn about the untapped potential of uncommercialized IP Most inventions don’t get commercialized. Investments to create patents and other forms of IP often are not recovered due to burdens of costly legal processes, enforcement issues and poor discoverability. Technologies including AI, industry standards, and blockchain are beginning to change the landscape, enabling greater exploitation of existing IP. As I learn about the issues, I’ve been experimenting with NotebookLM, an AI tool that can create an “original” podcast episode from just a little bit of text.  Here’s a link to the episode I “created” (did I?). It’s a bit redundant toward the end and I recommend listening on 1.5x speed, but if you have any interest in how quickly IP management is changing, you will find the first few minutes interesting.  https://lnkd.in/gXFJHRUk Here’s a link to the Big Idea, Inc. website where I “borrowed” the content (a straight copy and past into a document) that I loaded to NotebookLM. (If you haven’t tried this yet, give it a try. You’ll be blown away by how quickly it can produce a compelling new audio program from just a little bit of text).  https://lnkd.in/gzcmpVca Thank you to Nate H. at Big Idea for the content in his blog. I’m learning a lot about IP! Please comment if you are using NotebookLM. What do you use it for? Do you have any IP concerns about using AI? --- Created by a human (because I found the AI generated version of this post crappy).

11

3 Comments

IP Strategy for AI In the venture world, really the business world at large, AI seems to be table stakes to just be in the game. Given the very different nature of AI than most other tech to date, how can companies develop strong IP strategy to leverage its position in the market and protect their business? We'd love to hear your thoughts. We'll be having a conversation with expert IP strategist, David Kalow, AB'74, David Kalow, AB'74, JD'76 JD'76, who has extensive experience with a range of cutting edge tech, https://lnkd.in/eZeWGdR7. Please see the listing for his extensive bio. #AI #IP #AIstrategy #IPstrategy David Kalow Serval Ventures Katherine Dettra Sledge Smith VC Lab Decile Group Stanford University Graduate School of Business Stanford LEAD Stanford Institute for Human-Centered Artificial Intelligence (HAI) Rachel Chalmers

12

#Tokens aren’t inherently #securities just because they’re tokens. Plain and simple. While certain regulators may not share this view, the crypto community and judges who have opined on this topic generally agree. If all tokens are securities, it would follow that a non-security magically becomes a security by being represented by an alphanumeric string. Doesn't make very much sense. Tokens, however, can be the object of an investment contract, in which case they can be part of a securities transaction and must be analyzed to determine if they have the characteristics of a security. Similarly, tokens can reflect a digital representation of a traditional security—such as equity—in which case they are a security. This means you can't take equity, throw it on the blockchain, and claim it's not a security by virtue of being on the blockchain. That also wouldn't make much sense. There is a separation between a securities transaction of a securities instrument, such as a SAFT or TPA, and the subsequent token (if any). The SAFT/TPA could be a security, while the token could—and often does—represent something distinct, i.e., a utility. Meaning, a securities transaction involving a prospective token or digital asset does not automatically render the token or digital asset a security. It’s the contractual, transactional, or representational context that matters. We need to align on these key principles soon. Doing so will provide clarity, drive innovation, and create more opportunities for U.S. retail.

30

3 Comments

SB 1047 was a step toward complete AI regulation but failed due to its inherent problems where developers of models are burdened to implement requirements under Section 22603 (b). Overall, I believe the requirements would have created a barrier for companies to enter the AI space, and existing developers of models would have had to comply with high standards, which would come at a cost that not everyone can afford, thereby leading to the stifling of innovation. That being said, it's important that the legislators genuinely see the value technology has brought to the valley while promoting Responsible AI to achieve true AI Regulation. Feel free to share your thoughts—whether you agree or disagree.

26

There is a massive backlog of uncommercialized IP in America, and #TechTransfer offices play a critical role in licensing this IP to #startups and established businesses. The problem? It is an incredibly daunting undertaking, and to date, is largely a manual process. Enter #VentureIQ and its #TechTransfer abilities -- it can ingest an entire portfolio of patents and white papers (yes, you read that right!) and very quickly determine the market opportunity for the technology and identify businesses doing similar things who may stand to benefit (or compete) with the technology. Pretty cool, right?

5

The recent lawsuit filed against Nvidia and Microsoft over AI computing technology will impact how innovation in AI is protected under patent law. With companies pushing the boundaries of what’s possible in AI, this lawsuit could set precedents for patentability and enforcement, influencing future R&D investments and legal strategies in the tech industry. https://lnkd.in/g9j-SpDb #PatentLaw #AI #IntellectualProperty #TechIndustry #Innovation #LegalTrends

7

2 Comments

⚠️🚨 LLMs Will Always Hallucinate, and We Need to Live With This 🧐 👉 this paper provides detailed mathematical arguments why LLMs inherently hallucinate 👉 and why this function in current LLMs cannot be for 100 % mitigated 👉 outputs of LLMs, while still very useful in certain instances, therefore need to be carefully reviewed 👉 LLMs produce flows of text in a context but that output is not by design correct 👉 as a conclusion we might say that the output is related to the context input (e.g. of a prompt) by statistically likelihood (based on the systems training data and the respective algorithms applied) 👉 and in certain cases that is just was is needed (good enough) as an output #artificialintelligence

38

2 Comments

In early 2024, the biotech, AI, and cybersecurity sectors have seen an explosion of #megadeals, marking a notable surge in high-value investments. Here are some of the standout deals: Wiz: A cloud security startup raised $1 billion at a $12 billion valuation. This round was co-led by Andreessen Horowitz, Lightspeed Venture Partners, and Thrive Capital. Wiz plans to use the funds for further expansion and strategic acquisitions​. Xaira Therapeutics: This biotech firm secured more than $1 billion in a significant round led by ARCH Venture Partners and Foresite Capital. Xaira aims to leverage AI models to discover new drugs, a testament to the growing intersection of biotech and AI​​. Lambda: Specializing in cloud computing services and hardware for AI software training, Lambda raised $320 million in a Series C round, bringing its valuation to $1.5 billion. This funding round was led by the US Innovative Technology Fund​​. Pine Gate Renewables: This energy startup locked in $650 million to finance clean energy projects across the U.S., with investments from Generate Capital and other major backers​​. These deals highlight the robust investor interest and confidence in the potential of these sectors to drive future technological and healthcare advancements. For more detailed information, you can read the full article on Crunchbase here: https://lnkd.in/ewnKmjfz

13

1 Comment

🧐 Misinforming LLMs: vulnerabilities, challenges and opportunities 👉 a short paper on LLMs 👉 describes in a concise way how LLMs work (statistical models) 👉 and where the challenges are 👉 deals also with hallucination and issues related to misinformation 👉 I like the conclusion very much as it gets precisely to the point: In conclusion, the future of trustworthy LLMs, capable of generating justified statements based on given truth, dispute misinformation, and explain the self-reasoning process might require an hybrid combination of large transformer LLM, semantic web, and logic programming 🤓 #artificialintelligence #risk #future

20

The U.S. Department of the Treasury Issued the final rule on U.S. Investments in Certain National Security Technologies and Products in "Countries of Concern". The "Outbound Investment Security Program" will be administered by the newly created "Office of Global Transactions", within Treasury’s Office of Investment Security. The Final Rule will become effective on January 2, 2025. The technologies and products relevant to the prohibition and notification requirement are identified in the Outbound Order as semiconductors and microelectronics; quantum information technologies; and artificial intelligence. “US investments, including the intangible benefits like managerial assistance and access to investment and talent networks that often accompany such capital flows, must not be used to help countries of concern develop their military, intelligence, and cyber capabilities,” Paul Rosen, assistant Treasury secretary for investment security - #NationalSecurity #VentureCapital https://lnkd.in/eK3xzwXM

9

1 Comment