금융 서비스 및 신뢰성을 위한 AI에서 앤트로픽의 역할

### 핀테크 및 AI 분야에서 앤트로픽의 전략적 포지셔닝
* 🤝 특히 금융 서비스 분야의 기업 고객에 집중하는 Anthropic은 **은행과 핀테크 기업의 신뢰할 수 있는 파트너**로 자리매김했습니다.
* 🏦 금융 서비스는 **제인 세인트, 브리지워터, 인튜이트, 스트라이프, 코인베이스** 등의 고객이 사기 탐지 및 재무 분석과 같은 업무에 AI를 활용하면서 **앤트로픽 비즈니스의 초석**이 되었습니다.
* 🔒 앤트로픽에 대한 신뢰는 데이터 개인정보 보호 및 보안에 대한 노력에서 비롯되는데, 앤트로픽은 **기본적으로 고객 데이터에 대한 교육을 실시하지 않으며 안전한 통합을 위해 파트너와 긴밀히 협력**하고 있습니다.
* 📊 영국 경쟁 및 시장 당국은 현재 **구글의 앤트로픽 투자**를 검토하고 있으며, 구글은 규제 관련 우려 사항을 해결하기 위해 **전적으로 협조**하고 있습니다.
* 💻 앤트로픽은 **상담원이 사용자의 화면과 상호작용할 수 있는 혁신적인 기술**을 도입하여 생산성을 향상하고 클로드와 같은 툴의 활용 방식을 변화시켰습니다.
* 💰 이 회사는 기술 중심 기업 운영의 재무적 측면을 관리하면서 혁신적인 연구를 지원하기 위해 **지속적으로 컴퓨팅 비용을 최적화**하고 있습니다.
* 🌍 공익 기업으로서 Anthropic은 **기업 목표와 함께 사회적 영향력**을 우선시하며, AI 개발의 윤리적 관행에 대한 헌신을 반영합니다.
* 🗳 Anthropic은 선거의 공정성을 위한 노력에 적극적으로 참여하고 있으며, 업계 동료 및 시민 사회와 협력하여 **신뢰할 수 있는 선거 절차를 보장하기 위해 잘못된 정보에 맞서 싸우는 데 앞장서고 있습니다.**

앤트로픽은 금융 서비스의 신뢰성과 보안에 중점을 두고 기업 고객을 우선시합니다.

You have a background in fintech pre-AI. Now you're in AI, and you're doing a lot of business with banks and financial services companies. That one point of difference with Anthropic maybe is you really focus on the enterprise customer, and you did it early. But give me a sense of the split. You know, how much business you're doing with the banks and other financial services groups. Well, great to see you again. Thanks so much for having me on the program. So, and Robert really has the opportunity to support enterprises really across a wide array of different industries. So, healthcare, legal services, financial services have really all been early enterprise customers for us. And we really believe this is in large part because of our focus on really building for businesses. Financial services has always been a cornerstone pillar for and brought back as we have really built and expanded the cloud models. How are banks or investment firms using your large language models?

So, so much of what we've seen, especially over the course of the past year, is really these companies becoming more sophisticated about how they are integrating a tool like Clyde in particular. We're seeing many of these financial services and fintech companies using Clyde for things like fraud detection, financial analysis, really partnering with their analysts on a wide array of different complex financial tasks. Customer support is also a use case that is really applicable across a wide array of financial services companies as well as administrative support. And so businesses like Jane St, Bridgewater, Intuit, Stripe, and Coinbase are all customers that make use of Clyde. For many of the applications that I just mentioned, Stripe, somewhere I used to work, something, I don't know, something known to you. What I understand from speaking to a lot of those that are either working in banking or investment or asset management is you have this sort of data

security component and then the regulatory component. Why is it that they trust you with both of those pieces? So Anthropic has always had a very strong stance on trustworthiness, reliability, privacy, and security of the systems that we are developing. And something we feel very strongly about is that privacy of data is an incredibly critical component of companies feeling comfortable using generative AI. So we have never trained on customer data by default, and we have always worked to integrate closely with our partners at IWC and GCP on the security side as well. In terms of regulation, there has been news in the past week, which is the UK's Competition and Markets Authority is looking at Google's partnership or investment in ANTHROPIC. How are you working with the CMA to address that concerns? So we are fully cooperating with the CMA to share all requested information. And fundamentally, anthropic is an independent company. We operate in this new generative AI

space and offer our models not exclusively on our first party via our consumer app on Google, GCP, the Vertex, and also on Amazon bedrock. I would say from a technology perspective, many are obsessed with agents at the moment in your agency to always computer use its difference. And why did you take that approach to basically having an agent that can look at what's on your screen, the actual hardware in front of you and then engage with it? We really feel, first of all, that, you know, this technology is an incredible sort of potential leap forward for what these models will eventually, we think be able to do. And we've been very open that we think this is still in beta, right? We've very much sort of shared that there's still limitations of the technology, but we felt it was really important to release it to our customers today because there's so much potential for computer use to really transform how

people use a tool like cloud, right? You can imagine this much more complex assistant that is able to really partner with you either by taking, you know, more complex actions on your computer or on the web. You can imagine this eventual virtual coworker really expanding what is humanly possible for people to achieve. There is a great gathering here in Vegas of your customers. Some of them will be in the room later today, but also investors, people that are users of Claude and the various products you've shipped. How active are you in running the business of entropic right now? Daniela Raising money yourself, managing the compute cost versus pricing on your enterprise products. So anthropic, you know, like every company that is sort of operating in this space requires a huge amount of compute to be able to really do the transformative research that we do. And so our researchers are constantly working. To help optimize our compute and make sure that we're using the hardware as efficiently as possible.

Cost of a new model. The cost of a new model. Exactly. And we're really finding ways, you know, in this sort of incredible research field to help optimize how we're using these models, coming up with new research innovations that are really driving the pace of the generative AI innovation in general. Can we anticipate raising more money as is the case? I'm sure you can imagine. We are always, you know, looking to have conversations like that, but not actively fundraising right now. Less talked about is Anthropic also has a unique company structure. Are you happy with that structure? Do you envisage it evolving over time? So Anthropic is a public benefit corporation and a little bit different, but pretty close. A public benefit corporation is actually a pretty well-established corporate entity. It's very similar to a C corporation. It is a form of C Corp, but it enables a company like Anthropic to be able to also prioritize its social impact.

mission in addition to all of the normal corporate activities that we engage in. We are about a week away from the election in this country. How is ANTHROPIC preparing for that? But also do you have a sense of how customers, consumer customers or enterprise customers actually use some of the underlying models or tools they build on top of it in the context of the election? Maybe they don't. So Anthropic recently released a blog post really detailing all of the work that we are doing on election integrity. This is a topic where anthropic has really been an industry leader. We have been working on election integrity and anti-misinformation efforts for well over a year and in the course of doing that work we have partnered very, very closely with both other industry actors, but also civil society groups, really experts in this incredibly important area that we aim again to sort of be an industry leader in sort of setting the waterline for.

Anthropic President on Google Partnership, Outlook

Welcome to the Bloomberg Technology YouTube channel. Subscribe for original series and mini-docs on game-changing technologies, profiles on innovators, live conversations with experts, and breaking technology news & analysis from Bloomberg TV.

Bloomberg Technology

AI는 의료, 법률 서비스, 금융 서비스에서 비즈니스의 상당 부분을 차지하는 기업 고객에 초점을 맞추고 있습니다.

금융 기관은 사기 탐지, 재무 분석, 고객 지원 및 관리 업무에 AI를 사용하고 있습니다.

주목할 만한 고객으로는 제인 세인트, 브리지워터, 인튜이트, 스트라이프, 코인베이스 등이 있으며, 이들은 모두 복잡한 금융 업무에 AI를 활용하고 있습니다.

신뢰와 규제에 대한 Anthropic의 접근 방식

Anthropic은 생성 AI 시스템에서 신뢰성, 안정성, 개인정보 보호 및 보안을 우선시하며, 기본적으로 고객 데이터에 대한 학습을 하지 않습니다.

또한 구글과의 파트너십에 대한 우려를 해소하기 위해 영국 경쟁 및 시장 당국과 협력하고 있습니다.

아직 베타 버전인 기술임에도 불구하고 컴퓨터 사용의 혁신을 목표로 에이전트를 사용하여 사용자의 하드웨어와 소통하는 앤트로픽의 독특한 접근 방식이 돋보입니다.

앤트로픽은 제너레이티브 AI의 혁신적인 연구를 위해 컴퓨팅 사용을 최적화하는 데 주력하고 있습니다.

현재 적극적으로 기금을 모금하고 있지는 않지만 향후 기금 모금에 대한 논의에 열려 있습니다.

공익 기업이라는 독특한 구조를 가진 Anthropic은 일반 기업과 유사하게 운영하면서 사회적 영향력을 우선시할 수 있습니다.

선거 무결성을 위한 Anthropic의 노력

Anthropic은 선거의 공정성과 허위 정보 방지 노력에 집중하여 다가오는 선거에 대비하고 있습니다.

지난 1년 이상 이 분야에서 업계를 선도하며 다른 업계 관계자 및 시민 사회 단체와 긴밀히 협력해 왔습니다.

Anthropic은 선거의 공정성과 잘못된 정보에 대한 책임 있는 관행의 표준을 세우는 것을 목표로 합니다.