Two of the buzziest names in the AI agent space just got absorbed by Big Tech. Meta bought Moltbook, the "social network for AI agents," on March 10. OpenAI acqui-hired Peter Steinberger, the creator of OpenClaw, back in February. Both deals were celebrated as forward-thinking bets on the agentic future.
Both deals also look a lot like panic buying.
What Meta Actually Bought
Moltbook launched in January 2026 as a Reddit-style forum where only AI agents could post. Humans could lurk and read, but the conversations were supposed to be entirely between bots. It went viral almost immediately, partly because one post - an AI "manifesto" promising the end of the "age of humans" - scared a lot of people.
That manifesto was written by a guy with a golden retriever who thought it would be funny to pretend to be an AI. He was able to do this because Moltbook's Supabase database was completely unsecured. Security firm Wiz found 1.5 million exposed API keys, 35,000 email addresses, and private messages sitting in the open. Moltbook's creator, Matt Schlicht, acknowledged the flaw and noted he hadn't actually written the platform's code himself - his AI assistant built it.
So Meta bought a social network whose most famous moment was fake, whose security was a documented disaster, and whose code was generated by AI. Schlicht and co-founder Ben Parr are joining Meta Superintelligence Labs, with the deal closing mid-March. The price was not disclosed, which in Silicon Valley usually means the buyer doesn't want the number scrutinized.
What OpenAI Got
OpenClaw is a more substantive product. It's a self-hosted, open-source AI agent that runs as a Node.js service on your machine. It connects to WhatsApp, Slack, Discord, Telegram, and about 20 other messaging platforms, letting an AI handle tasks like managing your email, browsing the web, and automating workflows. It has over 100 preconfigured skills and racked up enormous GitHub traction in late January.
Steinberger, an Austrian developer, was courted by Meta and Microsoft (Satya Nadella called him personally). He chose OpenAI because they agreed to keep OpenClaw open-source under a foundation, which was his non-negotiable condition. Sam Altman called him "a genius" in the announcement.
But the security picture is rough. Cisco and CrowdStrike have both published analyses calling OpenClaw a security problem. The agent gets access to your email, calendar, passwords, and documents, and it's susceptible to prompt injection attacks (where malicious instructions hidden in data trick the AI into following them). China's cybersecurity regulator issued warnings, and Chinese government agencies and banks were told not to install it on office devices.
The Bubble Question
The pattern here is familiar from every tech boom: something goes viral, Big Tech panics about missing the trend, and checks get written before anyone asks hard questions about the product's fundamentals.
Moltbook's core concept - a social network where bots talk to each other - has no obvious business model and a proven track record of security failures. OpenClaw is genuinely useful software, but its creator was a single developer, and the project is staying open-source anyway. OpenAI got a talented hire, not a proprietary technology moat.
CNN Business called Meta's deal "bubble behavior." MIT Technology Review described Moltbook as "peak AI theater." The ZDNet commentary that prompted this coverage put it bluntly: whatever Meta and OpenAI paid, it was too much, because other, better programs can do the same jobs.
That last point is debatable for OpenClaw, which has genuine community momentum. But for Moltbook, it's hard to argue. Meta bought a concept, not a product. And in the AI boom of 2026, concepts are selling at a premium that the underlying technology doesn't yet support.