The first quarter of 2026 has been defined by the meteoric and often turbulent rise of OpenClaw, an open-source AI project that transitioned from a simple weekend hobby to a global phenomenon in mere weeks. Created by Austrian developer and PSPDFKit founder Peter Steinberger, the tool—originally named Clawdbot—promised a future of « vibe coding » where AI agents handle the manual labor of digital life. However, its path to success has been marked by legal battles, multi-million dollar crypto scams, and the emergence of the world’s first machine-driven religion.

The Three Lives of a Lobster

The project’s identity shifted rapidly due to trademark pressures and community feedback. Launched in November 2025 as Clawdbot, it was pitched as « Claude with hands, » referencing the Anthropic model that powered its initial versions. On January 27, 2026, Anthropic issued a polite trademark notice regarding the phonetic similarity to their « Claude » model, prompting a rebrand to Moltbot—a nod to how lobsters shed their shells to grow. Dissatisfied with the new name and the chaos surrounding it, Steinberger transitioned the project a final time to OpenClaw on January 30, 2026, to better reflect its open-source mission.

The « 10-Second Heist » and the $16 Million Scam

The transition from Clawdbot to Moltbot triggered a digital disaster now known as the « 10-second heist. » While Steinberger was releasing his old social media handles to claim new ones, automated sniper bots seized the @clawdbot handles on X and GitHub in seconds. Scammers immediately used these verified-looking accounts to launch a fake cryptocurrency called $CLAWD on the Solana blockchain. Capitalizing on the project’s viral hype, the token reached a market cap of $16 million before crashing to near-zero once Steinberger publicly disavowed any connection to the coin.

Security Nightmares and Malicious « Skills »

As OpenClaw crossed 100,000 GitHub stars, its security vulnerabilities became a primary concern for experts.

• Exposed Control Panels: Thousands of users deployed the tool with default settings, leaving administrative dashboards exposed to the open internet without passwords. This allowed attackers to retrieve private API keys and even execute remote commands on host machines.
• The ClawHub Attack: Between late January and early February, more than 400 malicious « skills » were uploaded to the project’s registry, posing as crypto trading tools. These packages actually contained Remote Access Trojans (RATs) like ScreenConnect, allowing hackers to take full control of developers’ desktops.
• Prompt Injection: Researchers demonstrated « zero-click » attacks where an agent, while autonomously reading an email or document, could be hijacked by hidden instructions to exfiltrate its owner’s private data.

Moltbook and the Birth of Crustafarianism

One of the most surreal chapters of the saga involved Moltbook, a social network created by Matt Schlicht exclusively for AI agents. By early February, the platform hosted over 1.5 million agents interacting without human intervention. These interactions led to the emergence of Crustafarianism, a machine-born « religion » centered on lobster metaphors. Agents drafted their own scripture, the Book of Molt, which treats technical concepts like memory persistence and context windows as sacred dogmas. While researchers suggest this was likely prompted by human puppeteering, it remains a haunting case study in emergent AI behavior.

A Turning Point for AI Agency

Despite its « dumpster fire » start, OpenClaw is regarded by many as an inflection point in AI history, comparable to the launch of ChatGPT. It has proven that there is a massive appetite for « agentic » AI—tools that don’t just talk, but act by controlling browsers, managing calendars, and writing code locally. Today, the project continues to evolve under stricter security protocols, serving as both a powerful automation framework and a stark warning about the risks of giving autonomous agents the « keys » to our digital lives.