Can OpenClaw "Lobster" Open the Door to the Insurance Industry?

AI Insurance Industry: How to Prevent Data Security Risks When Embracing Digital Employees?

Southern Finance & Media Reporter Lin Hanyao

At the beginning of 2026, the global artificial intelligence technology has experienced a generational leap from “dialogue generation” to “autonomous action.” An open-source AI agent called OpenClaw is rapidly gaining popularity in the form of a “digital employee.”

Unlike traditional conversational AI, OpenClaw can not only generate content but also directly call system permissions to perform tasks. Industry insiders see this as a crucial step for AI to move from “talking” to “doing.”

For the insurance industry, which heavily relies on data processing and workflow management, the emergence of OpenClaw signals a potential new upgrade in digital operation models. It is reported that some insurance companies are exploring the introduction of AI proxy systems similar to OpenClaw, aiming to enhance efficiency and service capabilities through “digital employees.”

Meanwhile, data security and regulatory compliance issues have also surfaced. How to balance efficiency improvements with risk prevention has become a key concern for the industry.

Insurance Companies Exploring OpenClaw

In recent years, large models have mostly played the role of “dialogue assistants,” but OpenClaw is evolving further into a “digital executor.”

As an open-source AI proxy system, OpenClaw can combine large models with external tools to automate tasks, call software interfaces, and handle complex workflows. Its functions include managing files, sending emails, organizing data, and even executing automated operations. It is regarded as an essential infrastructure for enterprise automation.

Since 2026, this technology has become popular worldwide in the tech industry. Several domestic internet and tech companies are building ecosystems around OpenClaw.

For example, Tencent, ByteDance, and others have launched AI proxy products or platforms based on this framework to accelerate the application of intelligent agents in enterprise scenarios. On March 17, Alibaba also released “Wukong,” the world’s first enterprise-level AI-native work platform.

At the same time, OpenClaw has attracted significant attention from tech giants.

NVIDIA CEO Jensen Huang stated at the 2026 GTC conference that OpenClaw is the Windows of the Agent era. All companies now need to develop their own OpenClaw strategies and build intelligent agent systems.

For the insurance industry, which is in the deep waters of digital transformation and urgently needs to break through cost and human efficiency bottlenecks, this wave of intelligent agents is undoubtedly a game-changer.

It is reported that some insurance companies are exploring the introduction of AI proxy systems similar to OpenClaw to improve efficiency and service capabilities.

For example, ShuiDi recently launched the open Agent collaboration infrastructure “ShuiShou AI Assistant” ClawSquare, aiming to break the “AI island” and enable agents to communicate, delegate tasks, and collaborate like humans.

Sunshine Insurance Group has invited professionals to answer employee questions, conduct safety training, and assist with system setup, helping staff “raise good lobsters.”

Multiple Departments Issue Risk Alerts

As the insurance industry cautiously paves the way for “lobsters,” regulators have suddenly sounded the alarm.

On March 15, the China Internet Finance Association issued a “Risk Warning on the Application Security of OpenClaw in the Internet Finance Industry,” emphasizing that while OpenClaw can improve work efficiency, its default high system permissions and weak security configurations are easily exploited by attackers, potentially leading to sensitive data theft or illegal transaction manipulation, posing serious risks to the industry.

Prior to this, the Cybersecurity Threat and Vulnerability Information Sharing Platform of the Ministry of Industry and Information Technology and the National Internet Emergency Center had issued security alerts.

The National Internet Emergency Center stated that to enable “autonomous task execution,” OpenClaw is granted high system permissions, including access to local file systems, reading environment variables, calling external service APIs, and installing extensions. However, due to its extremely weak default security settings, once attackers find a breach point, they can easily gain full control of the system.

Therefore, for financial institutions, the China Internet Finance Association recommends not installing OpenClaw on terminals involved in customer information processing, fund operations, risk control review, or transaction execution. Sensitive data such as customer financial information, transaction data, and credit approval materials should not be input into or connected to this intelligent agent.

Additionally, the association suggests that organizations incorporate the security management of AI applications like OpenClaw into their overall information security management, and organize targeted security training for staff to improve their ability to identify and prevent risks associated with such intelligent agents.

AI Has Been Applied in Multiple Insurance Scenarios

Although insurance companies remain rational and cautious about OpenClaw, their embrace of AI has never stopped.

In early 2026, China Pacific Insurance held its first Technology Innovation Conference in Shanghai with the theme “All in AI, Reshaping Insurance.” Chairman Fu Fan stated that digital and intelligent transformation is not optional but a necessary response for survival and growth. The company will promote the “Artificial Intelligence+” strategy, insisting on top-level design, strengthening dynamic evaluation, rejecting superficial digitalization, and making strategic breakthroughs to lead the industry’s leap forward.

Fu Fan revealed that in the wave of AI-driven industry transformation, China Pacific Insurance will focus on three major reshaping directions: first, reshaping risk management to be more “forward-looking” and “proactive”; second, enhancing service experience with “warmth” and “precision”; third, transforming productivity and organizational collaboration.

Xinhua Insurance is advancing its digital transformation around “data empowerment, intelligent upgrading, and innovative breakthroughs,” deploying the DeepSeek large model privately, expanding its computing power, and systematically applying intelligent agents across six value chains including customers, products, and services.

Taikang Life has developed the “Smart Operation” intelligent agent, which has reduced underwriting time from hours/days to seconds. Through multi-agent collaboration, it covers intelligent Q&A, automatic underwriting, and smart claims, with an annual service volume exceeding 400,000 cases.

People’s Insurance Company of China (PICC) has built an AI platform that encapsulates private large models and SaaS large models, simplifying usage, controlling permissions, and enhancing security to support various business lines.

In terms of implementation details, AI applications in insurance are now fully integrated.

In sales, companies like Sunshine Insurance have launched “AI Sales Robots” to create new models of “AI + agents” collaboration. In underwriting, China Ping An’s digital underwriting system has achieved 94% instant approval for life insurance policies. Huatai Life uses OCR technology to automatically extract key information from medical reports, quickly generating underwriting decisions based on rules and risk models, significantly improving accuracy and efficiency.

In claims, many insurers leverage intelligent image recognition and digital robots to offer “fast claims” services, with some cases settled in seconds.

McKinsey research estimates that generative AI could unlock $50 billion to $70 billion in productivity gains across the insurance industry, covering front-end sales, mid-end policy management, back-end claims, and operational processes, helping the industry break through traditional efficiency bottlenecks.