What could make a nearly 70-year-old intangible cultural heritage expert, a 60-year-old retired aviation technical engineer, a mother with a 2-year-old child in her arms, and a 9-year-old fourth-grade primary school student all come together to queue up? The answer is the free installation site of OpenClaw (commonly known as "lobster"), which has recently become popular in the technology circle.
Nearly a thousand developers and AI enthusiasts gathered at Tencent
On March 6, a long queue formed at the entrance of Tencent Company in Shenzhen: nearly a thousand developers and AI enthusiasts came to Tencent Building, and with the free assistance of Tencent Cloud engineers, they completed the cloud installation of OpenClaw, and collectively became "shrimp farmers on the cloud."
Why does AI technology, which seems so unfathomable to outsiders, attract users of different ages from various fields to rush to try it out? Behind the "shrimp farming" craze, how should we deal with the new problems brought about by the rapid implementation of AI technology?
What can "lobster" do?
According to the reporter’s understanding, OpenClaw (formerly known as Clawdbot/Moltbot) is an open source, Local-First AI Agent framework. Unlike various AI applications that can only chat on the web, OpenClaw is a super assistant that can take over the user's keyboard and mouse permissions. OpenClaw can run in the user's terminal and directly call system APIs to complete complex tasks.
Search*Tencent Cloud one-click deployment of openclaw*One-click direct access
"I must tell you responsibly: This thing is not an intern, nor is it a chatbot. It is a super employee who has the highest authority, does not sleep 24 hours a day, knows how to reverse crawl, can write his own code, and can learn new skills by himself." Some users left their own sigh after experiencing OpenClaw.
Previously, Cheetah Mobile CEO Fu Sheng shared a case of using "lobster" to achieve work through live broadcast: Fu Sheng said that after breaking his leg during skiing during the Spring Festival this year, he could only lie in bed every day and could not work normally. It coincided with the emergence of "lobster" and started the exploration of "lobster farming". After 14 days of "raising lobsters", "lobsters" gradually "raised" from a "novice" agent that didn't know how to check address books to a "team" of 8 agents. More importantly, the 8 "teams" can not only work automatically 7×24 hours, but also can iterate themselves. According to Fu Sheng, thanks to the unremitting efforts of this "team", Fu Sheng's official account has changed from updating less than 20 articles a year to updating daily. The topic chosen by "Lobster" himself even achieved the highest number of reads in the history of the account: one tweet received 1 million+ views, and Fu Sheng only found out that it was the masterpiece of "Lobster" after he woke up.
According to user sharing, it can be seen that in actual use, OpenClaw's capabilities can cover multiple scenarios in work and life. At work, OpenClaw can be transformed into an efficient office assistant: after accessing the mailbox, it can automatically organize and classify unread emails and mark their priorities, push processing reminders regularly every day, and say goodbye to the anxiety of hundreds of unread emails; it can synchronize schedules on different platforms, automatically check recent meeting arrangements, and provide real-time reminders when the time is approaching, so you no longer have to worry about missing important meetings; it can also be connected with various document tools to quickly convert PDF, Word, PPT and other files into Markdown format, and papers and data can be directly fed to AI for summary after being converted in seconds, greatly improving data processing efficiency. It can even connect to knowledge base tools such as Obsidian and Notion. Users only need to send a sentence or a picture, and the content can be automatically archived, tagged, and typeset, transforming knowledge management from a tedious operation of manual maintenance to "one sentence command, full process automation."
"Most people today actually work with computers. If the computer has the ability to judge, approve, write emails, and schedule - the computer can complete most of the work. What 'Lobster' does is to hand over a complete computer to the Agent. In the past, we regarded the Agent as software, and it was just one of many software. But you Think of the Agent as a person - it should not be a piece of software, it should be equipped with a computer. The code is still the same code, and the Agent is still the same Agent, but when you give it all the permissions of the computer, it really looks like a person. "Fu Sheng lamented to OpenClaw before. "One person + one lobster = one team. This is not the future, this is now."
Although "lobster" is popular, you need to be careful when "raising" it
At 10 a.m. on March 6, the first wave of more than 80 lobster enthusiasts had already started queuing up in front of the Tencent building in order to install OpenClaw on the cloud organized by Tencent for free assistance. By 11 o'clock, hundreds of reservation numbers had been issued. Some people come to Nanshan from Longgang and Longhua in Shenzhen, and some people come from cities such as Hong Kong and Hangzhou to learn shrimp farming across the city. There were both professional developers from all walks of life on site, as well as a large number of enthusiasts with non-professional backgrounds.
Shenzhen fourth grade students installing crayfish on site
Tencent Cloud stated that starting from January this year, Tencent Lighthouse Lighthouse has maintained an iteration rate of one version per day. From the launch of the OpenClaw one-click deployment template on January 28, to the exclusive adaptation of eight major mainstream IMs such as QQ and QQ on January 29, and the launch of the visual configuration panel and Skills capability management panel on February 3, Lighthouse continues to transform complex operations into a "zero code" experience. The configuration process that originally required manual modification has been transformed into visual forms and switches; high-quality plug-ins scattered in the global community have also been integrated into the official selection market to achieve "one-click installation and one-click activation." At present, the number of Tencent Lighthouse developers and the number of calling cores have exceeded historical peaks many times, and the number of shrimp farmers on OpenClaw Cloud has exceeded 100,000 and continues to rise.
Retired aviation technical engineer installing lobster on site
According to reporters, domestic technology companies including Byte and Baidu have launched similar OpenClaw cloud server deployment solutions. Large model companies from China are also hoping to use the popularity of OpenClaw to promote their products. Step Star stated that after being fully open sourced on March 4, the global popularity of its new generation base model Step 3.5 Flash has further increased, and the number of model calls has climbed to the first place in the world on OpenClaw. Base models from China such as MiniMax M2.5, Trinity Large Preview (free), Kimi K2.5, and Claude Sonnet 4.6 have also become important "guests" of OpenClaw.
However, as an AI product at the forefront of technology, is OpenClaw really suitable for all ages?
According to reporters, since January this year, many official security agencies have issued warnings one after another: the popular OpenClaw is becoming a "gold mine" in the eyes of hacker organizations. Security monitoring shows that more than 40,000 OpenClaw instances have been exposed to the public network, 63% of which have exploitable vulnerabilities, and more than 12,000 instances have been marked as fully remote controllable.
"If you want to run through this project, you have to follow the tutorial to install Node.js (_javascript runtime environment) and Git (an open source distributed version control system). For programmers, these two tools are equivalent to having two windows without anti-theft nets next to your thick security door. It becomes a weapon in the hands of hackers in an instant.” Some people in the industry said that installing OpenClaw itself has hidden security costs, and ordinary users have no need to use environment code. Opening a high-risk door in order to try OpenClaw is “completely inviting the wolf into the house.”
Recently, the Ministry of Industry and Information Technology's Network Security Threat and Vulnerability Information Sharing Platform (NVDB) has monitored and discovered that some instances of the OpenClaw open source AI agent have high security risks under default or improper configuration, which can easily lead to network attacks, information leakage and other security issues. The relevant risk warning pointed out that because OpenClaw has "blurred trust boundaries" during deployment and has the characteristics of continuous operation, independent decision-making, calling systems and external resources, in the absence of effective permission control, audit mechanism and security reinforcement, it may perform ultra-private operations due to instruction induction, configuration defects or malicious takeover, resulting in a series of security risks such as information leakage and system control. The Ministry of Industry and Information Technology’s Cybersecurity Threat and Vulnerability Information Sharing Platform recommends that when deploying and applying OpenClaw, relevant units and users should fully verify public network exposure, permission configuration and credential management, close unnecessary public network access, improve security mechanisms such as identity authentication, access control, data encryption and security audits, and continue to pay attention to official security announcements and reinforcement recommendations to prevent potential network security risks.
It is not difficult to see that although OpenClaw can bring a more cutting-edge technological experience through AI technology, for most users, "lobster farming" still needs to pay more attention to security risks. As Jeremy Turner, Vice President of Threat Intelligence and Research at SecurityScorecard puts it:“Learn to swim before you jump into the ocean.”