주요 보험사가 AI를 활용해 내부 프로세스를 운영하는 기업에 대한 사이버 보안 및 기타 보험 제공에서 점차 발을 빼고 있는 것으로 전해졌다.

보험 시장에서 고객의 AI 활용에 대한 일관된 대응 기준은 없지만, 많은 보험사가 사이버 보안 및 전문직 배상책임보험(E&O, Errors and Omissions 보험)에서 AI가 생성한 결과물과 관련된 청구에 대해 계약 체결을 조용히 거부하는 사례가 늘고 있다고 업계 관계자들은 전했다. 일부 보험사는 AI 관련 사고를 보장하기 위해 보험료를 크게 인상하는 방식으로 대응하고 있다.

보험사들과 협력하는 AI 개발 및 컨설팅 기업 코드스트랩의 CEO 코너 딕스는 “수십 개 보험사가 AI 관련 오류 보장에 대해 재검토에 나선 것으로 보인다”고 말했다.

이어 딕스는 “많은 보험사가 AI 결과물을 보장하는 데 부담을 느끼고 있다”며 “AI가 어떤 과정을 거쳐 결과를 도출했는지 추적하기 어렵기 때문”이라고 설명했다.

그는 “이러한 흐름은 사이버 보안과 E&O 전반에서 보험 보장 범위를 제외하는 형태로 나타나고 있다”며 “최근 등장한 다양한 바이브 코딩 기반 솔루션과 AI 시스템에는 본질적인 리스크가 내재돼 있지만, 전체 처리 과정을 완전히 들여다보기 어렵다”고 지적했다.

보험업계의 이러한 우려는 2025년 11월 파이낸셜타임스 보도를 통해 처음 드러났다. 당시 AIG, 그레이트 아메리칸 인슈어런스 그룹(Great American), WR버클리(W.R. Berkley) 등 주요 보험사는 챗봇과 AI 에이전트 등 AI 도구와 관련된 책임을 제외한 보험 상품을 제공하기 위해 미국 규제 당국에 신청서를 제출했다. 당시에는 향후 AI 관련 오류를 제외하기 위한 사전 대응으로 해석됐다.

그러나 현재는 다수 보험사가 실제로 AI 오류를 보험 약관에서 제외하는 방향으로 정책을 구체화하고 있다고 딕스는 밝혔다. 그는 “일부 보험사는 AI로 인한 비즈니스 중단과 책임에 대한 보장을 제한하거나 아예 중단하는 움직임을 보이고 있다”고 덧붙였다. 이어 “아이러니하게도 많은 보험사가 내부적으로는 AI 활용을 확대하고 있다”고 말했다.

딕스가 이끄는 코드스트랩은 AI 보험 시장 확대에 이해관계가 있는 기업으로, 자사 AI 코딩 플랫폼을 ‘추적 가능하고 보험 적용이 가능한 기술’로 홍보하고 있다. 다만 업계 다른 관계자들 역시 유사한 보험사들의 움직임을 확인하고 있는 것으로 전해졌다.

보험사, AI 보장 제외 확대

글로벌 보험사 NSI 인슈어런스 그룹의 재무 부문 책임자 제이슨 비샤라는 “얼마나 많은 보험사가 AI 워크로드 보험을 거부할지는 아직 불확실하지만, 일부 보험사는 이미 AI로 인한 비즈니스 혼란을 보장 대상에서 제외하는 보험 상품을 출시하고 있다”고 말했다.

이어 “보험사들의 리스크 수용 범위는 지속적으로 변화하고 있다”며 “AI와 관련해서는 아예 리스크 대상에서 제외하고 보험 견적 자체를 거부하는 사례도 나오고 있다”고 설명했다.

비샤라는 일부 보험사가 AI 결과물에 대한 보장을 거부하는 반면, 다른 보험사는 증가한 리스크를 반영해 보험료를 인상하고 있다고 전했다. 구체적인 인상 폭은 공개되지 않았지만 상당한 수준이라는 설명이다.

그는 “모든 기업은 보험에 가입하고 있으며, 동시에 대부분의 기업이 일정 수준 AI를 활용하고 있다”며 “이러한 상황에서 보험 약관에 AI 관련 책임 제외 조항이 포함되고, 보험사들이 이를 회피하는 경향이 나타나고 있는가에 대한 질문에는 ‘그렇다’고 답할 수 있다”고 밝혔다.

또한 보험사들은 AI 공급업체와 AI를 활용하는 일반 기업을 구분해 대응하고 있다. 많은 경우 보험사는 AI 기업에 대한 보장을 아예 거부하는 반면, 기술을 사용하는 기업에는 일부 예외 조항을 두는 방식으로 접근하고 있다.

비샤라는 “AI 관련 기업이거나 순수 AI 기업이라면 현재로서는 보험 가입이 거절될 가능성이 높다”고 말했다.

최근 몇 달 사이 보험사들은 고객의 AI 활용 방식에 대해 보다 구체적인 질문을 던지고 있다. 이는 사고 발생 가능성에 따른 리스크를 정밀하게 평가하기 위한 조치다. 다만 이러한 심사 강화는 결과적으로 기업이 AI 워크로드에 대한 보험을 가입하기 어렵게 만드는 요인으로 작용할 전망이다.

비샤라는 “현재 AI를 활용하는 기업이라면 ‘AI 정책은 무엇인지, 어떤 절차를 따르는지, 실제 비즈니스에서 어떻게 활용하고 있는지’와 같은 질문을 받고 있다”며 “보험 인수 담당자들 역시 ‘기업이 AI를 어떻게 활용하고 있는지’에 대해 매우 상세하게 확인하고 있다”고 설명했다.

보험 적용 기준 ‘재편’ 단계

MSP 기업 엔소노의 보험 부문 CTO 필 카레키는 일부 보험사가 AI 결과물 보장에서 물러나는 움직임을 보이고 있지만, 이것이 업계 전반의 흐름인지에 대해서는 신중한 입장을 보였다. 그는 보험사들이 보장 방식을 지속적으로 실험해 왔다고 설명했다.

카레키는 “보험사들은 보장 여부를 판단할 때 엄격하게 통제된 AI 도입과 실험적 프로젝트를 구분하려 하고 있다”고 말했다.

이어 “현재 AI는 통제된 생성형 AI와 자율형 AI로 양분되는 양상을 보인다”며 “이제는 단순히 ‘AI를 사용하고 있는가’가 아니라 ‘얼마나 통제되고 있는가, 어떻게 관리하고 있으며 얼마나 안전하게 운영되는가’가 핵심 질문이 됐다”고 설명했다.

그는 보험사들이 AI 워크로드 보장이 수익성이 있는지 판단하려는 단계에 있다고 덧붙였다. 명확한 의사결정 범위 내에서 운영되는 통제형 AI는 상대적으로 보험 적용이 가능하지만, 모니터링이 어렵고 롤백이 불가능한 실험적 시스템은 보장이 어렵다는 분석이다.

카레키는 “이는 단순한 후퇴라기보다 재정립 과정에 가깝다”며 “보험사들은 특정 상품이 시장성이 있는지 확인하기 위해 보장을 확대했다가, 결과를 분석한 뒤 다시 진입 여부를 결정하는 전략을 반복하기도 한다”고 말했다.

또한 AI 시스템의 보험 가능 여부는 개별 고객 환경에 따라 달라질 수 있다고 설명했다. 그는 “보험사들은 기본적으로 보험 사업을 포기하려는 것이 아니라, 해당 영역이 수익성이 있는지 판단하려는 것”이라며 “각 계약마다 ‘AI를 어떤 용도로 사용하는지, 어떻게 통제하는지, 어떤 리스크가 발생하는지’와 같은 질문이 뒤따르고 있다”고 전했다.

AI 개발 및 컨설팅 기업 코드스트랩 CTO 도리안 스마일리는 현재 AI 시스템의 신뢰성이 낮다는 점에서 보험사의 우려가 합리적이라고 평가했다. 그는 “이론적으로 AI 모델은 동일한 입력에 동일한 결과를 내야 하지만, 실제로는 같은 입력에서도 전혀 다른 결과가 나올 수 있다”며 “결과의 정확성을 스스로 검증하지 못하는 경우도 많다”고 지적했다.

또한 “대부분의 AI 모델은 귀납적 추론 능력이 부족하고 자체 검증도 어렵다”며 “그럼에도 많은 기업이 수백 개의 자율형 에이전트를 디지털 직원처럼 운영하려는 것은 비현실적”이라고 평가했다. 이어 “새로운 정보를 학습하지 못하고, 정보를 안정적으로 조회하지 못하며, 자신의 결과를 검증할 수 없는 사람을 채용하지는 않을 것”이라고 비판했다.

한편 NSI 인슈어런스 그룹의 제이슨 비샤라는 AI 보험 가입을 고려하는 IT 및 비즈니스 리더들에게 투명성을 강조했다. 그는 “AI 활용 방식을 충분히 공개하지 않을 경우, 사고 발생 시 보험금 지급이 거절될 수 있다”며 “보험사가 해당 리스크를 사전에 인지하지 못했다는 이유로 보장을 거부할 가능성이 있다”고 경고했다.
dl-ciokorea@foundryco.com

Several major insurance carriers have begun to back away from providing cybersecurity and other insurance to companies using AI to run internal processes, insiders say.

While there’s no standard response to customer use of AI in the insurance market, many carriers are now quietly declining to write policies for claims related to AI-generated outputs in cybersecurity and errors and omissions (E&O) coverage, these observers say. Other insurance carriers are jacking up prices to cover AI-related claims, they say.

Dozens of insurance carriers appear to be rethinking coverage for mistakes related to AI, says Connor Deeks, CEO of Codestrap, an AI development and consulting firm that works with insurance firms.

Many insurance companies aren’t comfortable with covering AI outputs because they can’t track the reasoning path the AI took to come up with a result, he says.

“That’s playing out downstream with insurance companies basically carving out coverage, whether that’s across cybersecurity or E&O,” he says. “All of these vibe-coded solutions and these AI systems that people have constructed have inherent risk baked into the cake now, and you can’t actually see the full process.”

The insurance carrier concerns about AI workloads first surfaced in November 2025, when Financial Times reported that three major carriers, AIG, Great American, and W.R. Berkley, filed requests with US regulators to offer insurance policies that exclude liabilities tied to AI tools such as chatbots and agents. At the time, those requests appeared to be preemptive moves to be allowed to exclude AI mistakes sometime in the future.

But now, many carriers seem to be moving forward with plans to exclude AI mistakes from policies, Deeks says. Several carriers he’s been in contact with are moving to limit or end coverage for AI-related business disruptions and liabilities, he adds. The irony is that many insurance carriers are embracing AI for their own internal purposes.

Deeks’ company has a vested interest in AI insurance coverage — Codestrap markets its AI coding platform as traceable and therefore insurable — but other industry insiders have also seen similar carrier decisions.

Carriers find exclusions

It’s still unclear how many carriers will refuse to insure AI workloads, but several carriers are now writing insurance policies that exempt coverage for AI-related business chaos, says Jason Bishara, financial practice leader at global carrier NSI Insurance Group.

“The risk appetite is changing among the carriers, and it’s always constantly evolving,” he says. “With regard to AI, there are carriers that are just removing it from their risk appetite and declining to quote altogether.”

While some carriers have declined to cover AI outputs, others are building in rate hikes to cover the increased risk, Bishara says. While he doesn’t have numbers on the extent of the rate hikes, they are significant, he adds.

“Every business has insurance, and every business now is using AI to some extent,” he adds. “Are you seeing those liabilities and exclusions within these policies and an aversion to it from the carriers? The answer is yes.”

Carriers are also treating AI vendors differently than AI users, he says. In many cases, carriers are declining to cover AI vendors altogether, while they carve out exceptions in policies against covering AI at companies using the technology.

“If you’re an AI-related company or specifically an AI company, there’s a good chance that you’ll get a declination at this point,” he adds.

In recent months, many carriers have been asking detailed questions about how customers are using AI to better understand the risk of insuring potential mishaps, he says. Ultimately, this increased scrutiny will make it more difficult for companies to buy insurance for AI workloads.

“For everybody leveraging AI right now, you’re seeing questions like, ‘What are your AI policies? What are your procedures? How are you leveraging AI within your business?’” Bishara adds. “We’re getting a lot of questions from the underwriters on, ‘How do you leverage AI within your business?’”

Coverage in flux

Phil Karecki, CTO for the insurance sector at managed services provider Ensono, also sees some carriers backing away from covering AI outputs, although he’s not sure whether it’s a major trend. Insurance carriers continuously experiment with how to provide coverage, he notes.

Carriers have tried to separate tightly governed AI deployments from more experimental projects when determining whether to provide coverage, he says.

“You’ve got this bifurcation of AI, the governed generative and the autonomous pieces,” he says. “It’s no longer, ‘Are you using AI?’ It’s asking, ‘Are you using governed AI? How are you governing it? How are you keeping it safe and secure?’”

Carriers have been trying to determine whether covering AI workloads can be profitable for them, Karecki adds. Governed AI tools operating in a bounded decision-making process will be more insurable, while experimental AI systems with no monitoring and no easy rollback will be difficult to cover, he notes.

“There’s a repositioning versus a pullback, and that’s very common to the industry, and they will at times open up coverage just to see if it’s this type of insurance that will sell,” he says. “They will assess the results and what needs to change so they can decide whether to re-enter this marketplace or abandon it completely.”

In some cases, whether an AI system is insurable may come down to circumstances at individual insurance customers. Carriers in general don’t want to get out of the business of providing insurance, Karecki says.

“What they’re working for right now is, ‘How do I make this profitable, and is this sector insurable?’” he says. “They make those decisions on every application regardless, but now, depending upon what they’re being asked to insure, the questions will follow. ‘What are you using AI for? How are you governing it? What risks does that introduce?’”

It makes sense that some carriers have begun to question whether to cover AI outputs, given the current level of unreliability of most AI systems, says Dorian Smiley, CTO at Codestrap.

“The math says these models should be deterministic, like given the same input, you should get the same output,” he says. “But you can get very different output from the same input, and they can’t know if the answer that they’re giving you is actually correct.”

In most cases, AI models lack inductive reason and can’t review their own work, but many organizations are talking about deploying hundreds of autonomous agents and treating them like digital employees, he notes.

“The idea that these agents are going to become employees, autonomous people working in your organization, is insane,” he says. “You would never hire a person that can’t learn new information, can’t reliably retrieve information, or check their own work.”

NSI’s Bishara has advice for IT and business leaders looking for insurance coverage for their AI workloads: Be honest about how they’re using AI. If they try to hide their AI risks, they risk having their claims rejected when something goes wrong, he says.

“If you don’t fully disclose these things appropriately in the way in which you’re functioning and operating, it could be utilized as an excuse to deny a claim at a later date,” he says. “You don’t want a carrier to come back and say, ‘We didn’t underwrite to that risk. We asked these questions, and you didn’t disclose it.’”

I spent years at Meta engineering large-scale systems for billions of users, delivering sub-second latency and five-nines (99.999%) uptime. When we started Outmarket AI, I brought that same lens: scalability, reliability, sustainability. Not buzzwords but real engineering.

Commercial insurance turned out to be a different planet. Some departments were still on pen and paper, going through manila folders. Others had systems built on COBOL, mainframes from the 80s to handle claims. Nobody wants to touch them because the guy who understood the code retired years ago and didn’t leave notes. Underwriters, brokers, marketing, customer reps — everyone going through thousand-page policy documents, making million-dollar calls for businesses. According to McKinsey’s State of AI research, 78% of organizations are using AI in at least one business function. Insurance has been slower to change the way it operates day to day.

We started building AI products for a few lines of commercial business: workers’ comp, general liability and property coverage to better understand all the pain points. Consider workers’ compensation, which in itself is a beast. A human has to analyze injury claims, workplace risk factors, OSHA reports, medical records, claims histories and state regulations that differ wildly. For general liability, one has to dig through premises risk, operations exposure, vendor agreements and similar hassles for property coverages. Meaning a single policy decision might need someone to pull together dozens of documents from different sources and spend more time on clerical work as opposed to the real deal

Within weeks, our first client wanted it for every other line of business. Not just one department, but the entire organization. The pattern repeated with every new client as they quickly realized the same AI infrastructure could transform how they handled all of their commercial policies. That moment crystallized something for the founding team. We weren’t building a feature, but instead building an AI-backed infrastructure and I knew exactly what that meant from my time engineering at scale.

The AI part wasn’t what kept me up at night. Large language models (LLMs) can handle dense insurance documents. That’s been proven. What worried me was everything underneath.

First, scale. How do we build something that grows with more clients? And scale by users per client? What about seasonality when commercial insurance policy renewals peak? Q4 is a mess. Traffic doesn’t grow linearly. It spikes ~10x.

Second, reliability. We started with one LLM provider. It worked fine, but what will happen when traffic spikes? Everyone’s slamming the same LLM. That’s a nightmare of hitting rate limits, token limits. What if third-party systems go down? We all have seen this in action when ChatGPT went down

Third, data isolation. No insurer would tolerate its proprietary underwriting data bleeding into a competitor’s context window. Every client needs their own guardrails.

So we weren’t just building a system. We were building a beast that can’t flinch under pressure, can’t go dark when a provider fails and can’t leak data between clients.

We attacked each problem head-on.

For isolation, we went single-tenant. Every client gets their own instance, their own database, their own AI context boundary. No shortcuts.

For reliability, we designed the load balancers of AI agents that look at everything and most importantly, latency, cost, accuracy needs, provider health and make a call in real time. If one provider is down, it is now obvious that the traffic has to shift.

This orchestration layer was the breakthrough that can scale infinitely now.

Why is insurance the ultimate stress test for AI infrastructure?

Think about a mid-sized restaurant chain buying commercial insurance. They need workers’ comp for kitchen staff, general liability for slip-and-fall incidents, property coverage for equipment, outdoor dining coverage, liquor liability, theft protection. Probably a dozen policies total. And these are all thousands of pages of dense legal language, exclusions, endorsements, coverage schedules and many more.

Before AI, someone had to read all of this manually. Risk managers spent weeks on it, sometimes months, comparing quotes from various carriers, hunting for gaps, trying to catch redundancies and all of this manually. The mental load was brutal and mistakes were inevitable. I have seen claims denied because of a coverage gap buried on page 847 that no one saw. The policy looked fine. The exclusion that mattered was hiding in plain sight. When that happens, insurers fall back on their errors and omissions coverage (E&O) to protect against mistakes made by their employees while reviewing insurance. That’s how broken the manual process is and can easily lead to millions of dollars in claims.

A typical policy bundle containing 2K pages can now be ingested in 10 to 15 seconds. Even though speed is a big win, what’s more exciting are things that were not possible before. Quotes from various carriers can now be compared side by side in real time. AI flagging gaps automatically before they turn into claim denial. Underwriters can type questions in plain English. “Does this cover water damage from a burst pipe in an unoccupied building during winter?” Answer with citations to gain more trust and confidence. No human can process with that speed and accuracy. The humans are now reviewers and decision-makers and not document processors.

Surviving seasonality: Engineering for 10x traffic spikes

Insurance has a brutal seasonality problem. Policy renewals cluster around year-end. As soon as Q4 hits, traffic is expected to spike by ~10x. An architecture that runs fine in March can collapse in December if we haven’t planned for it.

Three things kept me up. First, caching. LLM caching is not like a typical web caching. Take these two questions, for example:

1. “What’s my deductible for property damage?”
2. “How much do I pay out of pocket for building damage?”

Both are basically the same question. How do we recognize that and not waste compute power?

Second, scaling. When renewal season hits, the largest client might need 10x the capacity overnight, but I don’t want to pay for that capacity year-round.

Third, routing. Not every query to LLM needs the biggest and the best model. A simple policy lookup doesn’t need the same horsepower as a complex one. Sending everything to one model means simple queries wait behind heavy ones.

We tackled each one.

For caching, we have semantic matching algorithms at multiple levels.

1. At the embedding level: We cache vector representations so re-injection would re-use the same embeddings.
2. At the query level: We use locality-sensitive hashing to spot similar questions and serve cached responses. If a question is already answered, then a similar question can use the same response without burning the compute power twice.

For scaling, each worker process can auto-scale horizontally based on queue length and current latency in-place. The largest client might go from 4 workers to 40, then scale back as soon as traffic drops. The key here is that scaling can be reactive, but for seasonalities, it can be predictive. If client X’s renewal rush started October 15th last year, then we can technically pre-warm their infrastructure on October 10th this year.

For routing, we built a classifier that examines incoming requests and sends them to the right model. A simple lookup can use a small, fast model; however, a complex coverage analysis workflow can be routed to more sophisticated models. This can cut cost by about 40% and actually improve P95 latency because simple queries are not jammed behind complex ones.

Now let’s put this together and we see users getting sub-second responses irrespective of quiet Tuesdays or chaotic Decembers. That consistency is what turns AI to what people can use at scale.

AI hallucinations kill trust; domain knowledge fixes it

Large Language Models (LLMs) fail in ways that regular software does not. In traditional software engineering, a database either returns the right row or throws an error, but an LLM will always return plausible-sounding nonsense and any system will happily pass it downstream unless we build detection mechanisms. Research published in Nature has shown that detecting these “confabulations” (arbitrary and incorrect generations) requires measuring uncertainty about the meanings of responses, not the text alone.

The root cause depends on how all these models learn. General-purpose LLMs train on public data crawled from the internet. They’re capable of broad reasoning without any domain expertise. If we ask a general LLM about insurance policy structure, it will give a reasonable-sounding answer drawn from insurance data that exists in its training set, which may or may not reflect the actual terminology, coverage structures and regulatory requirements that clients operate on. In any insurance, a reasonable-sounding yet wrong answer can lead to denial of claims or even regulatory violations, leading to millions of dollars in losses

Research on fine-tuning LLMs for domain knowledge graph alignment has demonstrated that when models are tuned to domain knowledge, then it can perform multi-step inferences while minimizing hallucination. So we built out our own knowledge graph for insurance, which holds definitions of how the industry actually works. Coverage types, policy structures, regulations, carrier-specific terms, claims workflows, how everything connects. It took years of domain expertise to build it and we are still fine-tuning it every time we run into a weird edge case. What we found out was that when our models were fine-tuned against this custom graph, they stayed inside verified boundaries instead of inventing plausible-sounding answers from pre-trained public data.

In practice, this makes a huge difference. If a user asks for coverage exclusions, then the system no longer hallucinates. It uses a knowledge graph as a source of truth. Any missing knowledge in the graph means uncertainty rather than confabulating an answer.

No system is perfect, though. Even with the knowledge graph, things slip through. I call it hallucination tripwires, an automated check that can catch AI when it’s making stuff up.

Model claims a coverage limit that’s nowhere in the source document? Tripwire.

Model references a policy section that doesn’t exist? Tripwire.

Model pulls a number that’s way outside expected ranges for that policy type? Tripwire.

An ACM survey on LLM hallucinations categorizes hallucination detection techniques into two: factuality and faithfulness approaches. When a tripwire triggers, a smart system won’t just log an error and move on. It will fall back to a secondary model for verification purposes. And when that fails, it will escalate to a human for review, depending upon the severity and confidence scores.

Hallucination detection is one piece. The other is model drift. Models can get worse over time and shift away from training data and accuracy drops. We track this constantly, checking against human-verified samples. When we see the numbers trending down, we fine-tune or adjust our prompts. Observability isn’t a nice-to-have; it’s a must for enterprise applications to stay reliable and win clients’ trust.

Databricks popularized a concept called medallion architecture, where raw ingestion produces what we call “bronze” data, minimally processed, potentially messy. AI-driven normalization transforms this into “silver” data with consistent schemas and validated fields. Further enrichment and cross-referencing produce “gold” data that’s ready for downstream analytics and reporting. This tiering can help serve different use cases appropriately. Real-time policy queries can work against silver data, whereas regulatory reporting and actuarial analysis must work off of gold-tier data with full audit trails.

Engineering principles that made the difference

A few principles that stand out when I look back in time.

AI is an infrastructure. Treat it that way from day one. Don’t bolt on scalability later. The early decisions on single-tenant v/s multi-tenant, sync or async, one LLM provider or several will compound. Unwinding them later will be painful and expensive as it may eat up a good amount of engineering resources and time and even get new features to stand still for weeks to months.

Build for failures. Providers go down, models hallucinate and demands can go up at any time, especially when we least expect it, so build the fallback paths before entering panic mode.

Observability is not optional. In regular early-stage software, we can skip the fancy monitoring, but in later stages, especially with AI systems, we can not afford to do that. No observability will mean shipping a broken output and being blindfolded about it.

Commercial Insurance has built its traditional processes around human limits, especially reviewing speed and mental bandwidth. AI can lift those limits up if and only if the infrastructure holds up to its expectations reliably, at scale and under pressure.

The difference between an AI demo and an enterprise AI system is not the AI models but the backbone, the infrastructure that doesn’t flinch.

This article is published as part of the Foundry Expert Contributor Network.
Want to join?