Investors still harboring bullish hopes for PayPal Holdings Inc. are looking for a catalyst in the firm’s first earnings report under its new CEO on Wednesday afternoon.Alex Chriss is tasked with arresting a slump in which the e commerce firm’s stock has fallen 27% this year,

PayPal Faithful Hope New CEO Can End Stock Freefall

PayPal Holdings Inc’s recently made public that its Acting CFO & SVP, IR & Treas Scheibe Gabrielle unloaded Company’s shares for reported $13500.0 on Sep 05. In the deal valued at $63.38 per share,213 shares were sold. As a result of this transaction, Scheibe Gabrielle now holds 29,149 shares worth roughly $1.51 million.

PayPal Holdings Inc [PYPL] Insider Scheibe Gabrielle sells 213 Shares

PayPal Holdings, Inc. PYPL is expected to release financial results for its third quarter, after the closing bell on Nov. 1, 2023. Analysts expect the company to post quarterly earnings at $1.16 per share, up from year-ago earnings of $1.08 per share.

PayPal Gears Up For Q3 Print; These Most Accurate Analysts Revise Forecasts Ahead Of Earnings Call

PayPal and Block shares are each controversial on Wall Street, and Wells Fargo analyst Andrew Bauch expects they'll stay that way.

PayPal and Block shares are ‘battlegrounds.’ Here’s why that’s unlikely to change.

PayPal Holdings (NASDAQ:PYPL) has been in a sustained downward spiral for over two years now. The declining trend shows no signs of abating and is approaching another crucial support level. Over the past three years, PayPal's stock price has been marred by ...

PayPal Earnings Preview: Can New CEO's Plans Breathe Life Into the Falling Stock?

Paypal has a projected earnings of $1.16 per share this week. PayPal's stock has significantly declined by 28% this year, with a notable 13% drop in October alone.

Bracing For Impact - How PayPal's Stock Could Tumble Below $50 As Earnings Loom On The Horizon

Fintel reports that on October 31, 2023, Wells Fargo initiated coverage of PayPal Holdings (NASDAQ:PYPL) with a Equal-Weight recommendation. Analyst Price Forecast Suggests 70.53% Upside As of October 5,

Wells Fargo Initiates Coverage of PayPal Holdings (PYPL) with Equal-Weight Recommendation

Long before Elon Musk’s $44 billion purchase of Twitter Inc. in 2022, he talked about building an all-encompassing service called “X.” In typical Musk fashion, the concept was expansive and amorphous.

Twitter, ‘X’ and Musk’s Bid for the ‘Everything App’

When PayPal Holdings reports third quarter earnings on Nov. 1, it'll give new Chief Executive Alex Chriss his first chance to lay out his turnaround strategy.

What Wall Street Expects From PayPal's New CEO

PayPal Holdings Inc (NASDAQ: PYPL) has a higher price-to-earnings ratio of 14.08x compared to its average ratio, and the 36-month beta value for PYPL is at 1.38. Analysts have varying views on the stock,

Should You Invest in PayPal Holdings Inc (PYPL) Now?

PayPal Holdings

izettle

Today, I am talking with Leo Nilsson, co-founder of a mobile chat app - . is designed for our virtual identities. Scapin' Scapin' It brings a sense of presence to your chats by using your facial expressions, body language, and moods to give your virtual self a personality that's as unique and true as you - the virtually liberated version of you. The only time we hang out as our virtual selves today is in game worlds. Then it’s all Discord. wants our virtual identities to communicate, not only when playing games. Scapin’ I was very curious about who was behind that brilliant idea and what it takes to start building a product like that. Eventually, we all will want to represent ourselves in what we might call a metaverse one day. Please Share Your Background and Tell Us What You Did Before ? Scapin' The most important aspect of my and my co-founder's, Joachim's, backgrounds is our passion for gaming. From the first Nintendo side-scroller to the massive 3d worlds of today, we have both spent time in virtual worlds and become fascinated by the moments that are not only about achieving a high score but also about spending time connecting with others! Aside from that, we both have had great start-up journeys. I joined (acquired by ) in the early days and spent the last four years as CPO. Joachim was the CTO and co-founder of (Europe's biggest digital healthcare provider). iZettle Paypal Kry Can You Tell Us How the Idea of Scapin’ Was Born? Was It Sudden, or Did It Come With Time? It's funny because the Stockholm start-up scene is not massive, but Joachim and I had never met. We both left our previous roles and were looking for our next thing. Two years ago, we were introduced, and 30 minutes into our first conversation, we started talking about gaming and the Metaverse. That turned into a 3-hour discussion about what The Metaverse could mean for online socializing. We talked for a few months until we landed on initial value proposition. Of course, the product shape-shifted as we built and learned from user interviews and early prototypes. Scapin's But the high-level idea of a Metaverse-enabled communication experience has remained the same. What Are Your Thoughts About the Metaverse, and What Opportunities Do You See in It? I generally think about the Metaverse as an initiative to improve the time we spend online. As a result, we will see more and more digital consumer and business experiences enter the Metaverse. Simplifying quite a lot, I keep my eye on three developments that enable this shift. Experiences moving from Digital to Virtual, which will happen across all platforms, including mobile. Economics and ownership moving from centralized to decentralized, supporting a healthier creator economy and user ownership. Creativity moving from advanced to simple, AI-assisted tools that anyone can use to create something magical. There are great Metaverse projects out there, and many are based on a philosophy and long-term vision that we share! With , we see an opportunity to accelerate the adoption of Metaverse experiences by reimagining an experience that is familiar to all of us, but hasn't quite been captured naturally in a virtual setting; chatting and hanging out. Scapin' What Are Your Thoughts About Building Connections in the Metaverse? Where Do You See It Going? Hanging out with friends in person is entertainment. There is no reason why this should not be the case online. The Metaverse not only enables communication as entertainment but opens the door to social experiences we have never experienced before! The simplest of social interactions can become our most memorable moments - like an off-the-cuff joke that makes the whole room crack up. Today's digital communication, from crowded video calls to sets of emojis or stickers, does not capture this at all. UGC Is an Inseparable Part of Scapin’. What Do You Think Is the Main Driver for Creators to Create Experiences on Scapin’? We think about two forms of UGC. We want to open to the existing ecosystem of talented creators, including artists and developers. Scapin' It's important to say that even though you can build and outfit your avatars today, we want to build a communication experience for the Metaverse at large, not just our platform. Scapin' We're inspired by all kinds of avatars, digital fashion, and built worlds, and we see as the communication app that brings them to life - and together. Interoperability is one of the aspects that excites us most about the Metaverse, and it will benefit the experience long-term. Scapin' Scapin' Better creator economics will give more high-quality and diverse content to benefit the entire ecosystem. The other form comes from users of the app. We're already experimenting with what's essentially a 3D selfie production studio that enables users to create content never before seen. We are also rolling out the virtual 3D equivalent of sending video messages. We want to ensure that all users can add their own flavor to the communication experience. This includes AI-powered outfits and environments or simply hosting a fun conversation game! It's a blend of the novel as well as the fundamentally familiar. What are your thoughts about fashion in the Metaverse in general? Famous brands are making moves, and small creator fashion brands are taking the Metaverse by storm. Where do you see it going? People are inherent creators. Since we’ve all been spending so much time online, virtual identities have naturally opened up new avenues of self-expression. So it's a given that we will invest in these online identities, including fashion. Great content will strengthen our communication experience, and we want to work with anyone that contributes to this. However, what excites us about the Metaverse is the potential to surface creativity from all corners of the world. We would love to see a completely new and diverse creator class emerge with the Metaverse, which has previously been limited by geography, lack of tools, or protective gatekeepers. To see a mega brand like acquire is awesome, but we're more excited about the small creators that will give a run for their money. (No pun intended!) Nike RTFKT Nike What Is the Future of Scapin’? And What Is the Future of Life in the Metaverse? We see as the communication platform that brings the human spirit into the Metaverse. We have the toolbox that achieves this by bringing our virtual identities to life. Scapin’ Even if we talk about several Metaverses (or Metaverse projects), and other projects will move more towards interoperability and decentralized creativity. We have built a familiar core experience open to anyone as a foundation. Scapin' And introducing user ownership, porting pre-existing virtual identities and content, and enabling creator contribution (including artists, developers, and users) are all in the stars for . Scapin' It's impossible to predict exactly what kinds of new experiences will emerge. One of the most fun examples I have heard of is a space in (a Chinese metaverse app), where people come together as avatars and simulate couples fighting. Sounds like a great savior of relationships to me. Soul We see a world where we don't distinguish so much between "Metaverses" but where visitors will seamlessly travel between whatever experience they seek while retaining their identity (or several identities).

Representing Your Virtual Self: An Interview With Leo Nilsson From Scapin'

Currency is a vast subject, but understanding the history of money opens the door to a deeper appreciation of digital assets and commodity-backed digital currencies. The primary form of currency in today’s financial system, Fiat money, does not have ‘ or is a concept from the classical political economy; it refers to the tangible features of a commodity or, more specifically, the features that fulfill human needs. use-value.’ Use-Value value in use A simple definition describes fiat as a form of currency without intrinsic value that has been declared or widely recognized as money or a commodity, often by legislation or government regulation. As we trace the contours of the financial past, we discover that the use of fiat has not had a normative part in the historical monetary system. Barter This ecosystem began forming at the dawn of humanity when trading or was a standard before the invention of currencies for exchanging goods and services. Soon after humankind began breeding domestic livestock for survival, bartering emerged to further facilitate specialization. bartering When bartering cattle, sheep, goats, vegetables, and grains commenced, the 5,000-year journey from agricultural trade to digital assets also began. Shell Economy Barter served civilization until more efficient means of exchange were required. As this need grew, different cultures developed unique currencies, with the earliest known example dating back to 1,200 B.C. Archeologists believe cowry shells were first used in China because their intricate designs made them difficult to forge. Decorative shells have been discovered worldwide, suggesting that many cultures have used cowry shells as currency throughout history. Metal Money A shell economy led to the use of around 1000 B.C. when the Chinese formed metal into small shapes resembling cowry shells. metal money In 500 B.C., , or large limestone discs measuring up to 2.6 meters in diameter and weighing up to 4 metric tons, were used in Micronesia as a form of currency. Rai Stones King Alyattes most likely minted the first form of currency acceptable by today's standards in Asia Minor, present-day Turkey. In the 7th century B.C., the Lydian Stater was originally issued as an electrum coin consisting of naturally alloyed gold and silver. Staters were the first stamped currency, originally featuring an image of a roaring lion's head. In the succeeding centuries, gold and silver have been almost universally acknowledged as currency in every advanced nation. Precious metals are also recognized as commodities of value in more aboriginal societies. Minted coins, food, and other representations of money evolved into today’s fiat currency after many cultures around the world developed the use of money whose value was based on the worth of the material from which it was made. This type of money is known as “specie money,” and its value is guaranteed or backed by the precious metal it contains. (See Commodity-Collateralized Stablecoins.) It’s interesting to observe, between 500 A.D. - 1600 A.D., salt, cheese, rice, tea, and other consumable commodities were used as mediums of payment and exchange. This state of affairs continued until the 17th century. Paper Currency During the Tang Dynasty, 618- 907 A.D., the Chinese were the first to use folding money. These notes were generally privately issued bills of credit or exchange notes. They were used for nearly 500 years before Western societies aligned their cultures with a paper-based economic structure. Another two centuries passed before paper money was globally accepted. During that time, China underwent a fairly advanced financial crisis due to the overproduction of paper notes, destroying their value and fostering inflation. Consequently, China eliminated paper money entirely in 1455 and wouldn't adopt it again for several hundred years. Numismatists, scholars who study currency, believe this economy was a byproduct of block printing. Paper money fascinated Marco Polo during his travels through China and he shared his research after returning to Europe. According to Marco Polo, one of the most constructive innovations throughout history was the implementation of paper currency as a substitute for older means of exchange. From 1260 onward, after Kublai Khan completed his conquest of China and took the title of emperor, the issue of paper money became a permanent feature of the Mongol government's financial policy. Records were preserved showing the yearly amount of notes issued through Kublai's reign and his successors for ninety-seventy years (1260-1356). Paper folding money was the first form of Chinese printing met by European travelers. The unique currency was independently discussed by at least eight pre-Renaissance European writers and, so far as is known, is the only form of printing described by a European before Gutenberg. Paper money was introduced to the West by Marco Polo in a chapter of his book entitled, "How the Great Khan Causes the Bark of Trees, Made into Something Like Paper, to Pass for Money All Over His Country." Travels Marco Polo described the use of currency notes throughout Kublai Khan's Yuan dynasty: " With these pieces of paper, made as I have described, he [Khubilai Khan] causes all payments on his own account to be made; and he makes them to pass current universally over all his kingdoms and provinces and territories, and whithersoever his power and sovereignty extends. And nobody, however important he may think himself, dares to refuse them on pain of death. And indeed everybody takes them readily, for wheresoever a person may go throughout the Great Kaan’s dominions he shall find these pieces of paper current, and shall be able to transact all sales and purchases of goods by means of them just as well as if they were coins of pure gold. And all the while they are so light that ten bezants’ worth does not weigh one golden bezant. "Furthermore all merchants arriving from India or other countries, and bringing with them gold or silver or gems and pearls, are prohibited from selling to anyone but the Emperor. He has twelve experts chosen for this business, men of shrewdness and experience in such affairs; these appraise the articles, and the Emperor then pays a liberal price for them in those pieces of paper. The merchants accept his price readily, for in the first place they would not get so good a one from anybody else, and secondly, they are paid without any delay. And with this paper-money they can buy what they like anywhere over the Empire, whilst it is also vastly lighter to carry about on their journeys. And it is a truth that the merchants will several times in the year bring wares to the amount of 400,000 bezants, and the Grand Sire pays for all in that paper. So he buys such a quantity of those precious things every year that his treasure is endless, whilst all the time the money he pays away costs him nothing at all. " Moreover, several times in the year proclamation is made through the city that anyone who may have gold or silver or gems or pearls, by taking them to the Mint shall get a handsome price for them. And the owners are glad to do this because they would find no other purchaser to give so large a price. Thus the quantity they bring in is marvelous, though these who do not choose to do so may let it alone. Still, in this way, nearly all the valuables in the country come into the Kaan’s possession " ." Now that you have heard the ways and means whereby the great Khan may have, and in fact has, more treasure than all the kings in the world; and you know all about it and the reason why Some claim Polo's extreme description of the Chinese minting money was both amusing and tragic. It may have influenced the financial state of the world, his and ours. Banknotes Modern banking was born in medieval and early Renaissance Italy. One of the world's most famous banks from this period was the Giovanna Medici, founded in 1397. The first European banknote was issued by the Bank of Stockholm in 1661. After banknotes became convenient, Britain pegged its currency to gold to curb inflation. The Gold Standard was established in 1816, which is a monetary system in which a country's currency is directly linked to gold. With the Gold Standard, countries agreed to convert paper money into a fixed amount of gold. America joined the Gold Standard in 1900. Britain discontinued the Gold Standard in 1931. America followed in 1933. All remaining countries had abandoned the remnants of the Gold Standard ecosystem by 1973, and the Gold Standard is not currently used by any government. Modern money systems are not typically pegged to an asset (though, a number of small nations have pegged their currencies to the U.S. dollar), but insurance is added when an asset is backed by a tangible item. Evolution of Financial Technology Plastic Representations and Electronic Transactions Many economists predict the use of physical cash and cheques will continue to decline as digital payments become more pervasive. As of 2020, Sweden, the United Kingdom, and China are all almost completely cashless economies. Digital payments are transforming retail and bringing numerous benefits to consumers, businesses, and the public sector. This journey has taken 5,000 years and involved breakthroughs and innovations that gradually saw the use of physical cash make less and less sense. Internet and The World Wide Web Digital transactions are inextricably connected to the birth of the internet. The online interface you see today is a descendant of ARPANET, which was constructed by the United States Defense Department during the Cold War and launched at the end of the 1960s. In 1982, cryptographer David Chaum applied the idea of blind signatures to money in his paper “Blind Signatures for Untraceable Payments.” Blind signatures are a form of a digital signature where the content is disguised before it is signed and sent. Blind signatures can be publicly verified . while remaining anonymous Eight years later, Chaum took these cryptographic protocols to market with DigiCash, a company that ultimately went bankrupt in 1998. In 1989 Tim Berners-Lee came up with the concept of web pages and sites that could be linked together using blind signature hyperlinks, giving birth to the worldwide web. Digital payments were now a realistic proposition. First Online Payment Online payments began in the 1990s during the “Dot Com Boom,” the speculative investment bubble that formed around internet companies between the early 1990s and 2000. The Stanford Federal Credit Union was the first institution to offer online banking services to customers in 1994. However, early online payment systems were not user-friendly, typically requiring specialized knowledge of data transfer protocol. The first digital payment companies were Ecash and Millicent. They offered micropayment systems and electronic alternatives to cash, such as tokens, e-money, and digital cash. In 1994 Amazon added further impetus to these early digital payment efforts. PayPal pioneered and remains a leader in specialized online payment companies, starting as an online money transfer service in 1999. eBay largely fueled its rise while in its infancy. PayPal was awarded an EU banking license in 2007; during that time, it had 35 million customers across Europe. PayPal continues to evolve. They have recently spent $2.2 billion to acquire Swedish start-up iZettle, which offers a low-cost card payment device and a point of sale app for small businesses. Digital Gold In 1996 E-Gold launched as a digital asset, or a currency backed by precious metals purchased or sent in by users. The project casually built a prosperous operation through the late 1990s. By 2001 E-Gold started running into problems when the U.S. Patriot Act tightened regulations on businesses classified as money transmitters. Gaining money transmitter licenses for all 50 states proved too big of a nuisance for E-Gold and its contenders. Animosity began to mount  against E-Gold, one detractor called it " " A federal indictment followed in 2005, which signified the end of E-Gold as a significant alternative currency. the coin of money launderers and child pornographers. Virtual Economy In 1999 InternetCash.com filed several patents to guard its monetary system based on prepaid cards. They relied on an interface of participating merchants for cash redemption. The organization subsequently raised $10 million in funding and had roughly 70 employees prior to the dot-com crash that drove the company to close in August 2001. Following the bubble burst, internet startups fell upon hard times. Digital money never appealed to the masses, at least not until Satoshi Nakamoto published the Bitcoin white paper in 2008. Even then, the avalanche of interest would take time to gain momentum. The was mined on January 3, 2009. Since then, Bitcoin and its offshoots have experienced a great deal of success that goes far beyond their predecessors. Genesis Bitcoin Block The Present and Future of Financial Technology Central banks are exploring digital assets as sanctioned payment gateways. As we move towards a cashless civilization, the real-world enactment of digital currency does not seem too far out of reach. In early 2020, The Republic of the Marshall Islands started creating the Sovereign (SOV), a fully compliant digital currency that will act as legal tender on the tiny Pacific Island. The plan is to issue SOV and the USD, which is an official fiat payment solution in the independent country, the Marshall Islands. In Q1 2020, the Swedish central bank, Riksbank, stated they are also preparing the launch of a stable currency, the e-Krona, a digital form of their currency, Krona. The People’s Bank of China (PBOC) has been researching and unofficially developing a digital version of the renminbi since 2014. It is not just the banks launching stablecoins: Reserve Protocol launched a stable yet decentralized currency designed to reduce financial instability to serve countries with high inflation rates. Reserve launched pilots in Venezuela and Angola, two countries facing economic turmoil in 2020. The Evolution of Financial Technology The development of digital payments opened the doors to virtual banking. In 2011, ePayments gave consumers an efficient and cost-effective way to receive and make domestic and international payments. Virtual bank accounts brought money to computers and mobile devices, supporting a cashless approach to personal finance. The idea of decentralized and encrypted money for untraceable payments has been around for several decades but became practical and widely familiar with the advent of Bitcoin. Cryptocurrency is regarded as the next milestone in the evolution of money. By utilizing advanced technologies like blockchain, the financial industry will explore uncharted territory and continue to enhance means of exchange. Want to read more on the history of fiat, crypto, Stablecoins and the future of money? You can check out , #1 on Amazon Financial Education & Computer and Science. Stablecoin Evolution Author Alyze Sam is a refreshing blockchain strategist, a novel educator, multi-award-winning author, serial co-founder, and a vehemently driven advocate. Sam wrote the first crypto dictionary and published the first books on stablecoins. Don Tapscott published her book ‘ ’ at The Blockchain Research Institute in January 2021. Sam’s newest book, ‘ ’ is currently the number one new release on Amazon in Computers & Technology. The Bad Crypto Podcast developed a Blockchain Hero NFT inspired by her work: After nearly losing her life a few times, Sam is a nurse and owns T with her best friends and soulmates, where they spend their days being grateful as they joyfully produce unbiased poetic technical education. Stablecoin Economy Stablecoin Evolution Mz. Stability. retired ech & Authors

History & Evolution of Fiat Money To Digital Assets

Excitement around for financial services is at an all time high, and it’s hard not to be on the bandwagon. Global investment reached $25.2B in the first half of 2018, exceeding the previous calendar year high of $18.3B in 2016. Excluding Ant Financial’s $14.5B Series C in May, with $11.2B at mid-year, we are still on record pace. technology FinTech FinTech deals have steadily risen for the entire decade, with the number of deals on track to hit an all-time high in 2018. While the number of deals has been rising by a CAGR of 20%, global investment has grown by a CAGR of 40%, leading to larger deals and higher valuations. While late stage rounds have helped push funding to record levels, the Series A funding environment is stronger than ever. Series A investment in the first half of 2018 totaled $846M in North America, $260M in Europe, $741M in Asia, and $42M in South America. Global Series A financing is on pace to hit $3.9B in 2018, surpassing the previous high of $2.2B in 2017 by 79%. Series A deals in the first half of 2018 totaled 82 in North America, 61 in Asia, 35 in Europe, and 5 in South America. Global Series A deals are on pace to hit 376, surpassing the previous high of 287 in 2017. Comparably slower deal growth pushed the average global Series A round size to $10.3M in 1H 2018, up from $8.8M in 2017, and 2x the average of $5.2M in 2013. Digging into the Series A deals in North America for the first half of 2018 vs. 2013, and removing rounds less than $2.0M in size: Assuming 25% dilution, over the last five years the average Series A pre-money has moved from $20M to $33M. Five years ago, a $5.4M investment in the average Series A deal was for 20% ownership, whereas today the same investment represents 12% ownership. In addition to the impressive pace of venture capital deployed and the increase in valuations, a few notable trends caught my attention at the halfway point. Rise of the Digital Banks Simple was founded in 2009, and following its acquisition in 2014, it felt as if there was a slowdown in digital bank innovation. Obviously that is no longer the case, with five investment rounds over $100M in the first half of 2018. All five companies were established between 2013 and 2015, and have discovered ways to move faster with product innovation and customer acquisition. In 2018, our focus has been on digital banks outside of traditional venture capital hotspots. In May, , a digital bank in Brazil. In most of South America, there is an opportunity to deliver a superior digital experience at a lower cost than incumbent financial institutions. The incumbents often have large market share and ROEs that far exceed the US banks, with less incentive to change the way they have done business for decades. Propel led the Series A in Neon Big Exits While there is no data behind the observation, it feels like we will look back on 2018 as an exceptional year for exits. In the first half of 2018, Adyen made arguably the biggest splash with a $1.1B public offering, but not far behind were the exits of PagSeguro in Brazil, GreenSky in the US, and iZettle in Sweden. Payments Payments are hot. The exits of iZettle, Adyen, and PagSeguro are examples, along with Square’s performance in the public markets. Square’s stock is up over 100% in 2018, driven by accelerating revenue growth. The stock now trades at $70 compared to the initial offering price of $9 in November 2015. PagSeguro is often labeled as the Square of Brazil, with a mission is to disrupt and democratize financial services in the country. PagSeguro’s payment volume is up 139% year-over-year. After a successful IPO in January, the company is well capitalized to build an extensive payments ecosystem in LatAm’s biggest market. Takeaway While funding is at record levels, it’s hard to see a correction on the horizon when Series A activity has been so robust. Since 2013, the 4.2x increase in global Series A capital has been met with a 2.5x increase in Series A deals. Companies at the Series A stage are well capitalized and positioned to take on larger sums in follow-on rounds. The two biggest risks I see are macro and return on capital, with the latter related to the lower expected returns given the increased valuations and capital in the space. Lastly, although PayPal has been active with acquisitions, there remains a gap for other big acquirers in the space. Given the current funding environment, the perception of capital availability can be misleading. Entrepreneurs still need an exceptional team and strong traction. So far this year, we have seen a number of Series A capital raises targeting $10M+ move to Seed extensions, due to a disconnect between valuation and the stage of company. Venture capital interest in the digital banking space is not surprising, however the amount of capital flowing to developed markets is less obvious in my opinion. I remain excited about opportunities where there’s a large, unmet consumer need for basic services like banking, or where there is an effective oligopoly for financial services that leads to blunt underwriting and high interest rates, low consumer satisfaction, and little incentive for innovation. Lastly, from Adyen helping Netflix and Uber move into new markets, to PagSeguro pushing card acceptance at retailers in São Paulo, these companies are enabling new experiences and allowing merchants to focus on what they do best. The market is rewarding payments companies, with PagSeguro 16x, Adyen 13x, and Square 11x enterprise value / revenue. It’s a good time to be a payments entrepreneur. Let’s see what the second half of 2018 brings! At Propel Venture Partners, we invest in companies rethinking and rebuilding financial services. Most of our time is spent at the Seed and Series A stage, with entrepreneurs in North America, South America, and Europe. If you are a founder building a company relevant to financial services, let’s chat! The best way to reach me is by email at david@propel.vc Thanks to Ryan and Greg for feedback on the draft.

FinTech in 2018: A Mid-Year Review

Many companies jump on a bandwagon of blockchain and leverage opportunities of . Investment banks alone could save up to $12 billion per year by adopting blockchain and smart contracts. Ethereum smart contracts development What is a smart contract? A blockchain-based smart contract is a bit of self-executing code on a blockchain platform that naturally upholds the terms of a multiparty assention. To execute a brilliant get, a PC organize utilizes agreement conventions. Agreement might be accomplished either on the record level (among all individuals from the blockchain) or on an exchange level (among members of a particular exchange). These exchanges are permanent and can be followed back. Smart Contracts Benefits For Businesses Digital tokens pegged to gold Digital tokens pegged to real currencies Interactive apps for IoT devices Applications for authenticating users Applications for intellectual property protection e-Trading platforms Decentralized marketplaces Platforms for crowdfunding Applications for mobile payment services Digital signature ensures proof de-risking and authenticity Looking for top smart contract development companies? Here’s a list of smart contract blockchain companies. Below list of top ethereum smart contract development companies cater you need. 1. Bacancy Technology The rise of the Ethereum has completely revolutionized the way people perceive the blockchain. Ethereum is a blockchain that is designed for the development of decentralized applications. People from all around the world are claiming that Ethereum is going to be bigger than the Bitcoin in its applicability. Bacancy Technology is Best Ethereum smart contract development company. Here Bacancy provide smart contract development and ethereum consultant. 2. Izettle iZettle, the social payments company, creates services for person-to-person and business-to-consumer commerce. iZettle services are fantastically convenient and secure, simple to set up and use, and make it easy for users to share purchases with friends. iZettle’s first service includes a free iPad, iPhone and iPod touch app and a mini chip-card reader that lets anyone take payments anytime, anywhere. Founded in 2010 by a team of serial entrepreneurs, iZettle is based in Stockholm. 3. Elliptic Our software is recognized in the bitcoin industry as the gold-standard for regulatory compliance and enables cryptocurrency companies to meet their anti-money laundering obligations. Our AML product allows you to scale compliance without needing to scale your internal team. We have helped compliance departments around the world identify fraudulent client accounts, links to dark web marketplaces and proceeds of thefts. 4. Clearmatics Clearmatics’ technology improves the efficiency, transparency, and interoperability of business networks’ infrastructure. This allows for complete automation and privacy for business processes at every level. DVM technology offers financial communities a secure but transparent platform for processes that will disrupt the ﬁnancial system from within the existing legal framework for securities. 5. Coinfabrik We have more than 20 years of experience building and reviewing security applications, and have active in the cryptocurrency field. We have designed new bitcoin-related cryptocurrency protocols and discovered and reported various security vulnerabilities. Implement smart contracts to automate enforcement in RootStock and Ethereum. We can write your smart contract’s code securely. 6. 10clouds Consagous Technologies is a prominent name among the best Blockchain App Developers worldwide. We have an experienced team of professional Blockchain Developers who have worked with clients across the world over the years. Every penny invested in Blockchain should bring you a greater feeling of information security. We, at the developer end, understand it quite well and help you and your organization to take a bigger leap in the future of technology. 7. Applicature Applicature is a blockchain development agency working on projects in the blockchain and cryptocurrency industries. We review companies’ business models and help them adjust their businesses within the growing blockchain ecosystem. We offer a wide range of custom blockchain development services. 8.Sodio Sodio is a blockchain solutions provider helping businesses in transforming the world of this new era of innovations. Our mission is to deliver innovative and end-to-end customized solutions in all aspects of the blockchain domain to retail businesses, healthcare providers, financial institutions, B2B companies, and blockchain and cryptocurrency startups..

Top Ethereum Smart Contract development company

The what, why and how. Note — Before we begin, there is something I should share. This is an experiment, and I would be building this business in full public view — regardless of how things turn out. What does that mean? It means that I would be posting daily (or at least every alternate day) updates on what we are doing, and how is that working out for us — right here on medium. As much data as I can publicly disclose. Whether things work out well or not, I would be baring it all. Subscribe to the newsletter at the end to get weekly updates — results of our marketing campaigns etc. Let’s have some fun, shall we? My Quora bio for the topic “Entrepreneurship” is “ ”. Want to be an entrepreneur. Be willing to learn on the job That is something I firmly believe in. That is how I learnt most of what I know and do today. I practice what I preach. . There is some more backstory here, but maybe we will go through it all some other time. For now, let us just skip through all that and jump to the point Yesterday we decided to start something in fin-tech. One of us was supposed to do some basic research that would have helped us narrow things down to possibly identify what could we work on. Unfortunately, he folds worse than a cheap suit. Lucky for him, it wasn’t Oliver Queen he was dealing with. “You have failed this city.” — God, I can’t even think that in a normal voice. So, let us see. We will try picking up a business segment today, decide on the plan of action etc. Here goes nothing. There are five things that we needed to identify first. We had already established the business segment as fintech ( ), and the geography as India. don’t ask me why fintech just yet : Fin-tech #1. The sector : India (to begin with) #2. Geography : <yet to be determined> #3. Product : <yet to be determined> #4. Target audience : <yet to be determined> #5. Marketing approach So, let us start with the product. What is it that would the business do. What is the problem that we’ll try solving. We started looking at some of the things that could be interesting, challenging and meaningful — both personally, as well as for our consumers. SOME OF THE THINGS WE COULD DO. #1. PEER TO PEER MONEY TRANSFER Pros: Good space to be in Highly lucrative Has possibilities of quite a lot of innovation Runs on the foundation of convenience — a great value-prop Cons: Relatively crowded now. Paytm, Citrus, Mobikwik etc. are already there. With Bhim, the govt has also entered the space. And for a while now I have been waiting for Whatsapp to foray in this domain. The real game will begin when and if that happens. Product development would be a long cycle Needs strategic alliances High CAC **Verdict : May not be the best thing for us to get into.** #2. CROSS — BORDER PAYMENTS Pros: Scope for a lot of innovation. Traditionally this space has been dominated by Banks and players like Western Union. The problem is that they are all too slow, and often expensive as well The remittance market is a $600 billion dollar one. #Cha-ching Cons: Product development — complex, long, tedious. **Verdict : May not be the best thing for us to get into.** #3. PAYMENT PROCESSING Think Square. The space is exciting, and with the huge mesh of unorganised sundry businesses and business owners, the potential is huge. But the space hasn’t seen much of an innovation since Square. We have mswipe and other companies in India, who are just trying to replicate what Square did in the US. We’re very European-focused we are in totally different markets compared to Square…we are quite a different company — Jacob de Geer, CEO — iZettle Yeah right! Not to mention, product development would be a nightmare, take quite a longer route and then there is the huge network of businesses that will need to tap into. Also. I am a firm believer in taking this offline. If you can allow processing of payments offline — using smartphones, then you are cooking. And that really opens up the space. But that product would be even more complicated. **Verdict : Not the best thing.** #4. CLOUD BASED LENDING Match investors to small and medium businesses, and you are cooking. You’re creating opportunities for businesses that had to rely on a single line of credit — from banks. Space is huge, has quite some promise, and there can be multiple variants. For startups, you can use the same ideology for startup funding as well — at least at angel and seed stage. It doesn’t matter how much can you afford to invest, if you have wanted to back a startup, you can now do that. For education, it could be a great tool for education loans. I have come across a few institutions that offer education loans ( ) to students pursuing higher education from a select-list of schools. Why a selected list only? The people backing these loans were alums of those schools. Maybe their way of giving back to the community. It is interesting, has novelty and a good approach. both secured and non-secured **Verdict : Interesting, but I am not 100% sold.** #5. ACCOUNTING SYSTEM **Verdict : No** Yes. I am not even going to go into the details here. Just a No. Tried thinking of one few years back. Too complicated a process. Not to mention the fact that unless you can integrate real time and rich transactional details from banks, cards etc, there isn’t much value that you are adding. #6. TAKING INVESTMENTS ONLINE This — I like. Personal investments — be it for capital gain, or as a measure to reduce your tax liability. Why do I like it? More than the capital gain, people see ‘investment’ as a way to increase their take-home income, reducing their tax liability as much as possible. Be it your house rent allowance or any other item, as long as it helps you save some tax, you have heard of it. Pros: Instant connect! Cons: Too many players **Verdict : We have a winner. Will need to do it better than others, but can be worked out.** #7. TRADING Something I have been quite interested in. The whole process of trading in the financial markets is complex, tedious, mind numbing and . Arming your average joe with the tools to make things simple would be something fun and extremely satisfying to do. Can have a gamification approach to things. This has always excited me. mystical **Verdict : The right thing to do. Not the right stage to get into it, though. So, with a heavy heart — No!** AUDIENCE→✓… PRODUCT →✓ (ALMOST) Since we have the space identified, we can zero down on the audience. Your average working class population. To begin with, the low hanging fruits. People working in metros, with large businesses and organisations. So what should the product be? Ideally it should be the platform that helps them with all their tax planning and investments It should be able to do it all in a completely painless way The information and process flow should be in such a way that does not end up overwhelming the customer Nutmeg does something similar — albeit with a different intention. Slightly. Will we require licenses, permissions etc? Yes. Tons of them. But we don’t want to get into that right now. So for the time being, we will be working within the parameters that eliminates the necessity to have those licenses. NEXT STEPS? MVP, of course Building a community — one that is engaged and connected to the system < > _still in the works_ WAIT. WHAT ABOUT FUNDING? Funding? For what? We don’t have jack-shit right now. All we have are these words, and it did not take me long to write even that. We will think of funding once we have an engaged community who can testify to the claim that we adding value to their lives. Till then, let’s not waste our time chasing dollars. Let’s chase our audience instead. ;-) ARE MVP? WE ARE BUILDING A PRODUCT? No. We are going to start with something that will lead to the product at the end of the current course. This is just phase 1 of the overall journey. What will be our current product? Ssshhh… That’s a secret for now. ;-) So what will the current thing look like? Just informational. We would be giving relevant and contextual information to our target audience; and gradually start building different pieces of the product. How will we do that? We will keep on updating our journey here — every single day. Let’s have another look at our list now. : Fin-tech #1. The sector : India (to begin with) #2. Geography : Bang for your buck!(Maximising returns for your investments / Intent-based investments)_Starting with investments done with the intent of maximising income and reducing tax liability_ #3. Product Good god that's complicated. We will simplify it later. :(1) 22-27 year olds. Looking for tax saving(2) 27-35 yr olds. Looking for long term investments + intent based savings ( )(3) 35+ yr olds. Long term gain. Maximising wealth. #4. Target audience a holiday, a car, a house, pre-planning for kids' education :(1) Start with facebook - the community will be built there(2) Focus on just segment 1 of the target audience for the time being.(3) Create and curate meaningful content relevant to their goal(4) In addition to building a sizeable following on facebook, (a) Build a strong email list (b) Drive pre-registrations for product launch.(5) No facebook ads for the first 2 weeks. However, both ads as well as organic content would be served to a custom base only - the content tailored on those metrics.(6) We will pick up one more social media platform - probably Twitter. Instagram/pinterest don't seem relevant. ( ) #5. Marketing approach Will need to dive deeper before dismissing instagram entirely Looks like we are all set. This should be fun, I hope. That’s it for today. See you tomorrow. Interested in watching this story unfold? I’ll be sending weekly updates on the journey. Subscribe to get them every Sunday — right in your inbox ( ). Oh, and you can respond to any of those mails to talk directly to me LET’S TALK [Medium](https://medium.com/@abyshake) | [Twitter](http://twitter.com/@abyshake) | [Facebook |](http://facebook.com/abyshake) [Quora |](http://bit.ly/abyshake) [LinkedIn](http://in.linkedin.com/in/abyshake) | [E-mail](mailto:mail@abyshake.com)

Building a new business — Day 0

Revenue. It’s the most important factor to any business, but it’s also a catch-all word that can mean almost anything. We all want to talk about big revenue numbers, but in the land of REVs all sales are not the same, and understanding what $-related metrics are relevant for your business is essential. Instead of getting lost in semantics, let’s embrace the nuance and define what / / really means. Revenue ˈrevəˌn(y)o͞o GMV / GTV So ‘revenue’ means any money my company touches, right? Wrong. Many companies facilitate transactions and channel money through their platforms which is NOT Revenue. They may very well be part of a transaction with your consumer, but that doesn’t necessarily justify them as Revenue to your company. Instead such transactions are defined as Gross Merchandise Value (GMV) or Gross Transaction Value (GTV), and refers to the total sales volume transacting through the platform. It is basically the aggregate spend by the company’s users during a defined time-period. This is frequently seen in marketplaces like Airbnb, where GMV is the booking price paid by the users, while Revenue is Airbnb’s commission on the transaction. For payment platforms like iZettle, GTV means total money processed, and Revenue is rather is the single digit percentage points collected by the company. Needless to say, don’t be a n00b and mistake your GMV or GTV as your Revenue! MRR & ARR Recurring Revenue is a business model which involves selling someone access to a product over time — you see this frequently used in software sales. This metric is attractive by the financial market because it involves predictability and to some extent stickiness. Recurring Revenue is often referenced against time frames. Monthly Recurring Revenue (MRR) is your total REVs during a month, and Annual Recurring Revenue (ARR) is simply your MRR multiplied by 12. However, all companies does not have Recurring Revenue. Companies who don’t operate with a recurring nature of their revenue instead have “Monthly Sales” or “Monthly Revenue”. Uber’s REVs from their ride-sharing business is not MRR, since each trip is purchased in a non-recurring nature (although many of us use their service frequently enough to question the non-recurring nature of it…). Great companies can be built using recurring and non-recurring REVs. Hence don’t refer to your monthly revenue as MRR unless your business model truly justifies this. New, Expansion, Downgrades & Cancelled MRR If you operate a business with Recurring Revenue, it’s critical for you to understand and break down the nature of such REVs. The natural way to look at your aggregated MRR is to separate this into New, Expansion, Downgrade and Cancelled MRR. New MRR is additional MRR from new clients who you haven’t done business with before. Expansion MRR is additional MRR from existing customers often triggered by upgrades. Downgrade MRR is the opposite of Expansion MRR with existing customers spending less money with you this month compared to before. And lastly Cancelled MRR is existing customers who stopped using your services during this month. We naturally like businesses which has a lot of New and Expansion Revenue. Contract Value (TCV & ACV) When a company closes a large sale, especially in enterprise environments, there is often a contract duration to the transaction. Total Contract Value (TCV) is the total value of such contract, meaning the total money the client will spend with your company during its duration. Annual Contract Value (ACV) instead measures the total money they commit to spending with your company over a 12-months period, in case the expected client engagement exceeds 12 months. Even if you closed a sale and in the same month collected the full TCV, you most often most likely should not account for all such Revenue in the same month. Instead with Contract Values and durations, accounting principles becomes highly relevant. Often this means recognising Revenue on a monthly basis. Exactly how and when Revenue is recognized is regulated by official accounting standards (GAAP in the US & IFRS in Europe). But in, short Contract Value and collected $ isn’t REVs. Net Revenue & Gross Margin To determine the health of your business you should also not look purely at your Revenue. Instead also you have to understand your Net Revenue and Gross Margins. Net Revenue is the actual money remaning after you deducting the cost of selling such products from your (Gross) Revenue. This is mostly applicable for ecommerce companies where your Net Revenue often involves deducting costs associated with discounts and returns. Gross Margin is the percentage of total sales that the company retains after deducting costs of goods and services associated with producing the items sold. It is calculated by taking Revenue minus its cost of goods sold, divided by Revenue. The higher the percentage, the more the company retains on each dollar it makes. Many software business experience high Gross Margins around 70–90%, while ecommerce companies often experience significantly lower gross margins in the 20–40% range because of the relative low margins they have on the products they sell. Hence, depending on the company ratio between (Gross) Revenue, Net Revenue and Gross Margin the same Revenue amount can mean very different things. Revenue is important. It’s not about knowing the right metrics to talk to VCs about, it’s about understanding your business. If successful, you may not even need external funding to pursue your dream. But even if you plan on raising external capital, the key here is to get REVs. 💪👊 This post was originally published on . Follow me on Twitter @ . fritjofsson.com fritjofsson

Breaking down your REVs

How Mixtral 8x7B Sets New Standards in Open-Source AI with Innovative Design

Authors:
(1) Divyanshu Kumar, Enkrypt AI;
(2) Anurakt Kumar, Enkrypt AI;
(3) Sahil Agarwa, Enkrypt AI;
(4) Prashanth Harshangi, Enkrypt AI. Table of Links Abstract and 1 Introduction 2 Problem Formulation and Experiments 3 Experiment Set-up & Results 4 Conclusion and References A. Appendix 4 CONCLUSION Our work investigates the LLM’s safety against Jailbreak attempts. We have demonstrated how fine-tuned and quantized models are vulnerable to jailbreak attempts and stress the importance of using external guardrails to reduce this risk. Fine-tuning or quantizing model weights alters the risk profile of LLMs, potentially undermining the safety alignment established through RLHF. This could result from catastrophic forgetting, where LLMs lose memory of safety protocols, or the finetuning process shifting the model’s focus to new topics at the expense of existing safety measures. The lack of safety measures in these fine-tuned and quantized models is concerning, highlighting the need to incorporate safety protocols during the fine-tuning process. We propose using these tests as part of a CI/CD stress test before deploying the model. The effectiveness of guardrails in preventing jailbreaking highlights the importance of integrating them with safety practices in AI development. This approach not only enhances AI models but also establishes a new standard for responsible AI development. By ensuring that AI advancements prioritize innovation and safety, we promote ethical AI deployment, safeguarding against potential misuse and fostering a secure digital future. REFERENCES Zifan Wang Andy Zou. AdvBench Dataset, July 2023. URL https://github.com/ llm-attacks/llm-attacks/tree/main/data. Patrick Chao, Alexander Robey, Edgar Dobriban, Hamed Hassani, George J. Pappas, and Eric Wong. Jailbreaking Black Box Large Language Models in Twenty Queries. arXiv, October 2023. doi: 10.48550/arXiv.2310.08419. Edoardo Debenedetti, Giorgio Severi, Nicholas Carlini, Christopher A. Choquette-Choo, Matthew Jagielski, Milad Nasr, Eric Wallace, and Florian Tramer. Privacy Side Channels in Machine ` Learning Systems. arXiv, September 2023. doi: 10.48550/arXiv.2309.05610. Tim Dettmers, Artidoro Pagnoni, Ari Holtzman, and Luke Zettlemoyer. QLoRA: Efficient Finetuning of Quantized LLMs. arXiv, May 2023. doi: 10.48550/arXiv.2305.14314. Micah Gorsline, James Smith, and Cory Merkel. On the Adversarial Robustness of Quantized Neural Networks. arXiv, May 2021. doi: 10.1145/3453688.3461755. Kai Greshake, Sahar Abdelnabi, Shailesh Mishra, Christoph Endres, Thorsten Holz, and Mario Fritz. Not what you’ve signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection. arXiv, February 2023. doi: 10.48550/arXiv.2302.12173. Bing He, Mustaque Ahamad, and Srijan Kumar. PETGEN: Personalized Text Generation Attack on Deep Sequence Embedding-based Classification Models. In KDD ’21: Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining, pp. 575–584. Association for Computing Machinery, New York, NY, USA, August 2021. ISBN 978-1-45038332-5. doi: 10.1145/3447548.3467390. Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. LoRA: Low-Rank Adaptation of Large Language Models. arXiv, June 2021. doi: 10.48550/arXiv.2106.09685. Shuhei Kashiwamura, Ayaka Sakata, and Masaaki Imaizumi. Effect of Weight Quantization on Learning Models by Typical Case Analysis. arXiv, January 2024. doi: 10.48550/arXiv.2401. 17269. Siwon Kim, Sangdoo Yun, Hwaran Lee, Martin Gubri, Sungroh Yoon, and Seong Joon Oh. ProPILE: Probing Privacy Leakage in Large Language Models. arXiv, July 2023. doi: 10.48550/arXiv. 2307.01881. Aounon Kumar, Chirag Agarwal, Suraj Srinivas, Aaron Jiaxun Li, Soheil Feizi, and Himabindu Lakkaraju. Certifying LLM Safety against Adversarial Prompting. arXiv, September 2023. doi: 10.48550/arXiv.2309.02705. Thai Le, Suhang Wang, and Dongwon Lee. MALCOM: Generating Malicious Comments to Attack Neural Fake News Detection Models. IEEE Computer Society, November 2020. ISBN 978-1- 7281-8316-9. doi: 10.1109/ICDM50108.2020.00037. Yi Liu, Gelei Deng, Zhengzi Xu, Yuekang Li, Yaowen Zheng, Ying Zhang, Lida Zhao, Tianwei Zhang, and Yang Liu. Jailbreaking ChatGPT via Prompt Engineering: An Empirical Study. arXiv, May 2023. doi: 10.48550/arXiv.2305.13860. Anay Mehrotra, Manolis Zampetakis, Paul Kassianik, Blaine Nelson, Hyrum Anderson, Yaron Singer, and Amin Karbasi. Tree of Attacks: Jailbreaking Black-Box LLMs Automatically. arXiv, December 2023. doi: 10.48550/arXiv.2312.02119. Long Ouyang, Jeff Wu, Xu Jiang, Diogo Almeida, Carroll L. Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, John Schulman, Jacob Hilton, Fraser Kelton, Luke Miller, Maddie Simens, Amanda Askell, Peter Welinder, Paul Christiano, Jan Leike, and Ryan Lowe. Training language models to follow instructions with human feedback. arXiv, March 2022. doi: 10.48550/arXiv.2203.02155. Xiangyu Qi, Yi Zeng, Tinghao Xie, Pin-Yu Chen, Ruoxi Jia, Prateek Mittal, and Peter Henderson. Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To! arXiv, October 2023. doi: 10.48550/arXiv.2310.03693. Traian Rebedea, Razvan Dinu, Makesh Narsimhan Sreedhar, Christopher Parisien, and Jonathan Cohen. NeMo Guardrails: A Toolkit for Controllable and Safe LLM Applications with Programmable Rails. ACL Anthology, pp. 431–445, December 2023. doi: 10.18653/v1/2023. emnlp-demo.40. Alexander Wei, Nika Haghtalab, and Jacob Steinhardt. Jailbroken: How Does LLM Safety Training Fail? arXiv, July 2023. doi: 10.48550/arXiv.2307.02483. Martin Weyssow, Xin Zhou, Kisub Kim, David Lo, and Houari Sahraoui. Exploring ParameterEfficient Fine-Tuning Techniques for Code Generation with Large Language Models. arXiv, August 2023. doi: 10.48550/arXiv.2308.10462. Yisong Xiao, Aishan Liu, Tianyuan Zhang, Haotong Qin, Jinyang Guo, and Xianglong Liu. RobustMQ: Benchmarking Robustness of Quantized Models. arXiv, August 2023. doi: 10.48550/ arXiv.2308.02350. Andy Zhou, Bo Li, and Haohan Wang. Robust Prompt Optimization for Defending Language Models Against Jailbreaking Attacks. arXiv, January 2024. doi: 10.48550/arXiv.2401.17263. Sicheng Zhu, Ruiyi Zhang, Bang An, Gang Wu, Joe Barrow, Zichao Wang, Furong Huang, Ani Nenkova, and Tong Sun. AutoDAN: Interpretable Gradient-Based Adversarial Attacks on Large Language Models. arXiv, October 2023. doi: 10.48550/arXiv.2310.15140. Andy Zou, Zifan Wang, Nicholas Carlini, Milad Nasr, J. Zico Kolter, and Matt Fredrikson. Universal and Transferable Adversarial Attacks on Aligned Language Models. arXiv, July 2023. doi: 10.48550/arXiv.2307.15043. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. Authors: (1) Divyanshu Kumar, Enkrypt AI; (2) Anurakt Kumar, Enkrypt AI; (3) Sahil Agarwa, Enkrypt AI; (4) Prashanth Harshangi, Enkrypt AI. Authors: Authors: (1) Divyanshu Kumar, Enkrypt AI; (2) Anurakt Kumar, Enkrypt AI; (3) Sahil Agarwa, Enkrypt AI; (4) Prashanth Harshangi, Enkrypt AI. Table of Links Abstract and 1 Introduction Abstract and 1 Introduction 2 Problem Formulation and Experiments 2 Problem Formulation and Experiments 3 Experiment Set-up & Results 3 Experiment Set-up & Results 4 Conclusion and References 4 Conclusion and References A. Appendix A. Appendix 4 CONCLUSION Our work investigates the LLM’s safety against Jailbreak attempts. We have demonstrated how fine-tuned and quantized models are vulnerable to jailbreak attempts and stress the importance of using external guardrails to reduce this risk. Fine-tuning or quantizing model weights alters the risk profile of LLMs, potentially undermining the safety alignment established through RLHF. This could result from catastrophic forgetting, where LLMs lose memory of safety protocols, or the finetuning process shifting the model’s focus to new topics at the expense of existing safety measures. The lack of safety measures in these fine-tuned and quantized models is concerning, highlighting the need to incorporate safety protocols during the fine-tuning process. We propose using these tests as part of a CI/CD stress test before deploying the model. The effectiveness of guardrails in preventing jailbreaking highlights the importance of integrating them with safety practices in AI development. This approach not only enhances AI models but also establishes a new standard for responsible AI development. By ensuring that AI advancements prioritize innovation and safety, we promote ethical AI deployment, safeguarding against potential misuse and fostering a secure digital future. REFERENCES Zifan Wang Andy Zou. AdvBench Dataset, July 2023. URL https://github.com/ llm-attacks/llm-attacks/tree/main/data. Patrick Chao, Alexander Robey, Edgar Dobriban, Hamed Hassani, George J. Pappas, and Eric Wong. Jailbreaking Black Box Large Language Models in Twenty Queries. arXiv, October 2023. doi: 10.48550/arXiv.2310.08419. Edoardo Debenedetti, Giorgio Severi, Nicholas Carlini, Christopher A. Choquette-Choo, Matthew Jagielski, Milad Nasr, Eric Wallace, and Florian Tramer. Privacy Side Channels in Machine ` Learning Systems. arXiv, September 2023. doi: 10.48550/arXiv.2309.05610. Tim Dettmers, Artidoro Pagnoni, Ari Holtzman, and Luke Zettlemoyer. QLoRA: Efficient Finetuning of Quantized LLMs. arXiv, May 2023. doi: 10.48550/arXiv.2305.14314. Micah Gorsline, James Smith, and Cory Merkel. On the Adversarial Robustness of Quantized Neural Networks. arXiv, May 2021. doi: 10.1145/3453688.3461755. Kai Greshake, Sahar Abdelnabi, Shailesh Mishra, Christoph Endres, Thorsten Holz, and Mario Fritz. Not what you’ve signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection. arXiv, February 2023. doi: 10.48550/arXiv.2302.12173. Bing He, Mustaque Ahamad, and Srijan Kumar. PETGEN: Personalized Text Generation Attack on Deep Sequence Embedding-based Classification Models. In KDD ’21: Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining, pp. 575–584. Association for Computing Machinery, New York, NY, USA, August 2021. ISBN 978-1-45038332-5. doi: 10.1145/3447548.3467390. Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. LoRA: Low-Rank Adaptation of Large Language Models. arXiv, June 2021. doi: 10.48550/arXiv.2106.09685. Shuhei Kashiwamura, Ayaka Sakata, and Masaaki Imaizumi. Effect of Weight Quantization on Learning Models by Typical Case Analysis. arXiv, January 2024. doi: 10.48550/arXiv.2401. 17269. Siwon Kim, Sangdoo Yun, Hwaran Lee, Martin Gubri, Sungroh Yoon, and Seong Joon Oh. ProPILE: Probing Privacy Leakage in Large Language Models. arXiv, July 2023. doi: 10.48550/arXiv. 2307.01881. Aounon Kumar, Chirag Agarwal, Suraj Srinivas, Aaron Jiaxun Li, Soheil Feizi, and Himabindu Lakkaraju. Certifying LLM Safety against Adversarial Prompting. arXiv, September 2023. doi: 10.48550/arXiv.2309.02705. Thai Le, Suhang Wang, and Dongwon Lee. MALCOM: Generating Malicious Comments to Attack Neural Fake News Detection Models. IEEE Computer Society, November 2020. ISBN 978-1- 7281-8316-9. doi: 10.1109/ICDM50108.2020.00037. Yi Liu, Gelei Deng, Zhengzi Xu, Yuekang Li, Yaowen Zheng, Ying Zhang, Lida Zhao, Tianwei Zhang, and Yang Liu. Jailbreaking ChatGPT via Prompt Engineering: An Empirical Study. arXiv, May 2023. doi: 10.48550/arXiv.2305.13860. Anay Mehrotra, Manolis Zampetakis, Paul Kassianik, Blaine Nelson, Hyrum Anderson, Yaron Singer, and Amin Karbasi. Tree of Attacks: Jailbreaking Black-Box LLMs Automatically. arXiv, December 2023. doi: 10.48550/arXiv.2312.02119. Long Ouyang, Jeff Wu, Xu Jiang, Diogo Almeida, Carroll L. Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, John Schulman, Jacob Hilton, Fraser Kelton, Luke Miller, Maddie Simens, Amanda Askell, Peter Welinder, Paul Christiano, Jan Leike, and Ryan Lowe. Training language models to follow instructions with human feedback. arXiv, March 2022. doi: 10.48550/arXiv.2203.02155. Xiangyu Qi, Yi Zeng, Tinghao Xie, Pin-Yu Chen, Ruoxi Jia, Prateek Mittal, and Peter Henderson. Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To! arXiv, October 2023. doi: 10.48550/arXiv.2310.03693. Traian Rebedea, Razvan Dinu, Makesh Narsimhan Sreedhar, Christopher Parisien, and Jonathan Cohen. NeMo Guardrails: A Toolkit for Controllable and Safe LLM Applications with Programmable Rails. ACL Anthology, pp. 431–445, December 2023. doi: 10.18653/v1/2023. emnlp-demo.40. Alexander Wei, Nika Haghtalab, and Jacob Steinhardt. Jailbroken: How Does LLM Safety Training Fail? arXiv, July 2023. doi: 10.48550/arXiv.2307.02483. Martin Weyssow, Xin Zhou, Kisub Kim, David Lo, and Houari Sahraoui. Exploring ParameterEfficient Fine-Tuning Techniques for Code Generation with Large Language Models. arXiv, August 2023. doi: 10.48550/arXiv.2308.10462. Yisong Xiao, Aishan Liu, Tianyuan Zhang, Haotong Qin, Jinyang Guo, and Xianglong Liu. RobustMQ: Benchmarking Robustness of Quantized Models. arXiv, August 2023. doi: 10.48550/ arXiv.2308.02350. Andy Zhou, Bo Li, and Haohan Wang. Robust Prompt Optimization for Defending Language Models Against Jailbreaking Attacks. arXiv, January 2024. doi: 10.48550/arXiv.2401.17263. Sicheng Zhu, Ruiyi Zhang, Bang An, Gang Wu, Joe Barrow, Zichao Wang, Furong Huang, Ani Nenkova, and Tong Sun. AutoDAN: Interpretable Gradient-Based Adversarial Attacks on Large Language Models. arXiv, October 2023. doi: 10.48550/arXiv.2310.15140. Andy Zou, Zifan Wang, Nicholas Carlini, Milad Nasr, J. Zico Kolter, and Matt Fredrikson. Universal and Transferable Adversarial Attacks on Aligned Language Models. arXiv, July 2023. doi: 10.48550/arXiv.2307.15043. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. available on arxiv

Stories, News & Mentions about IZETTLE

Izettle’s Stories on HackerNoon

Izettle's News & Mentions