Creator of the viral OpenClaw in Sam Altman's team. This is how the AI market is changing
This is a hit transfer Peter Steinberg. It became famous last November when the world saw it Clawdbot (currently OpenClaw) – a free tool often described as personal AI assistant.
Its phenomenon is that thanks to access to the computer's memory and user data, it can be easily assigned to perform specific, often multi-stage tasks. For example, you can connect it to WhatsApp, Slack and your mailbox, from WhatsApp you can send it an order to summarize a presentation saved on your computer, send it by e-mail and inform the team about the completion of the task in Slack.
If it cannot connect to an application or service, you can simply ask it to “learn” it and download appropriate extensions created by users. Alternatively, creating a new one from scratch thanks to the advanced programming capabilities of the latest generations of language models.
The possibilities of this assistant are limited almost exclusively by your imagination (and the size of the wallet, more on that later) and in the eyes of many people this is a foretaste of what the future of AI looks like.
The explosion of popularity of this tool did not go unnoticed by the most important players on the market. One of them turned out to be Sam Altman himself, who announced in a post on X that Peter Steinberg joined the OpenAI teamto “power the next generation of personal agents,” and OpenClaw will now be developed under his company's guidance.
Objective? An AI agent that 'even my mother can use'
Peter Steinberg also wrote more about this on his blog. He mentioned, among other things, that his next mission is to “create an agent that even his mother can use”. Achieving this will require thinking through “how to do it safely” and access to “the latest models and research.”
The mention of security is certainly not accidental. OpenClaw, despite its many advantages, has a huge disadvantage: It gives cybersecurity specialists a headache. Non-technical people who don't know what they're doing often share their most confidential data with the world while using it, and various types of hackers keep coming up with new ways to obtain and use it.
If even our mothers are to use AI agents in the future, it will certainly be necessary to take care to prevent them from doing great financial, image or even legal harm using such an assistant.
Read also: The biggest internet lie of recent days. In Moltbook, people impersonate bots
OpenClaw, or from a viral curiosity to an important element of the OpenAI strategy
Hiring Steinberg and bringing the OpenClaw project under the umbrella is a very logical decision by OpenAI. Although ChatGPT is undoubtedly the most popular AI chatbot, in the rapidly developing world of AI assistants and agents, the position of Sam Altman's team is not that strong, and AI agents are perceived as the next step in the development of artificial intelligence.
In the development community, Anthropic model-based tools – especially Claude Code – have been praised for months for their exceptional agent-based performance, especially in scenarios requiring multi-step tooling and code manipulation. This is where some of the most enthusiastic discussion about “hands-on agents” has been centered in recent months. This narrative was further strengthened by the mentioned one Clawdbot, which was initially very closely associated with Anthropic solutionswhich resulted in an increase in their popularity.
If Anthropic acquired the creator of OpenClaw and his tool, many people would certainly become convinced that if they run AI agents, they should run them with Anthropic. Instead of cooperation, however, letters from lawyers appeared, which forced Steinberg to change the name of his work from Clawdbot, first to MoltBot, and then to OpenClaw.
OpenClaw can do a lot in the right hands.
Altman himself apparently saw this as an opportunity for OpenAI. In the race for agents, perception in the development environment is of strategic importance, because this is where tools and ecosystems are created and later reach mass recipients. By hiring Steinberg and taking over the development of the OpenClaw project has a chance to slow down the growth of the competition's popularity and gains a very important weapon in the fight for position on the market of next-generation AI tools, which in 2026 we will hear about in almost every business presentation.
Tools like OpenClaw may have it too of great importance from the perspective of the OpenAI business model. It is already known that OpenAI's enormous ambitions cannot be financed only by subscriptions paid by ordinary consumers.
AI agents or assistants operate on a different principle than popular AI chatbots, because in the case of more complex tasks, they are billed depending on the processed tokens. You can think of tokens a bit like watts, i.e. units of work performed by AI. The more difficult the task, the harder the AI assistant has to work and the more tokens generated by the accelerator it will consume. Simple tasks that you can wait for can be performed locally, on a small language model running on your computer. However, if the task is more complex and time is of the essence, it must be performed on accelerators operating in data centers and then you can quickly spend a lot of money on tokens.
There are plenty of examples on the Internet of people who spent thousands of dollars on tokens in a few days using Claude Code or OpenClaw.. Of course, these are extreme examples, but it is not difficult to see that multi-stage tasks assigned to AI agents can quickly become another important source of revenue for OpenAI, next to advertising, and OpenClaw has shown that there is no shortage of people ready to open their wallets if they feel that specific work has been done for them.
