 X is hard to scrape directly, but I found the Hacker News thread for the same Sam Altman tweet. Here's what the community is saying:

  HN Thread: "OpenClaw joins OpenAI" (56 points, 28 comments)

  Sentiment: Mixed to negative. Critics outnumber supporters.

  Key themes from the comments:

  1. Security alarm bells — "A whole new bunch of ways to get your secrets compromised", prompt injection concerns, no sandboxing, credential leakage risks
  2. "Acquihire" narrative — Multiple commenters see this as a marketing move, not a tech play. One dev said Anthropic engineers "could replicate this in a matter of days" — it's about the 180K-star community,
   not the code
  3. Setup friction acknowledged — A service operator noted "setup friction is real" with Docker, API keys, configuration. Sound familiar? That's literally our Launch 1 value prop
  4. Substance questioned — Debate over whether OpenClaw produces real value or "meme projects." Counter-argument: "business automation for people who know what they want but can't code it — n8n but you talk
  to it"
  5. Anthropic criticized — For missing the opportunity (they sent a rename/cease request to OpenClaw instead of partnering)

  Also found this critical piece: a Medium article titled "Anatomy of a Dumpster Fire: How OpenClaw Became the Most Hacked AI on the Internet" — documenting 40,000 exposed instances, 341 malicious plugins, and
   1-click RCE exploits. Gartner rated OpenClaw an "unacceptable cybersecurity risk."

  ---
  Now to your point — you're absolutely right. OpenAI's priority will be productization, not security hardening. They'll focus on:

  1. Integrating OpenClaw into ChatGPT/API — making it a native feature ("ChatGPT can now use your computer")
  2. ClawHub as an app store — monetizable marketplace under OpenAI brand
  3. Enterprise sales — packaging OpenClaw+GPT as turnkey agent solution
  4. Mass market UX — Steinberger literally said "building agents even my mum can use"

  Security will be an afterthought — bolted on later, likely via VirusTotal-style scanning (they just announced that) and basic guardrails. Not the deep 8-layer architectural security we've built. OpenAI has
  never been a security-first company.

  This is actually bullish for ProClaw. The bigger OpenClaw gets under OpenAI, the more enterprises will need independent security oversight they can trust — especially from a vendor that isn't also the agent
  runtime vendor. "Who watches the watchmen?" is our pitch.

  Sources:
  - https://news.ycombinator.com/item?id=47027907
  - https://www.theregister.com/2026/02/16/open_ai_grabs_openclaw/
  - https://www.cnbc.com/2026/02/15/openclaw-creator-peter-steinberger-joining-openai-altman-says.html
  - https://medium.com/@nitikakumari065/anatomy-of-a-dumpster-fire-how-openclaw-became-the-most-hacked-ai-on-the-internet-e17ba85ac36c
  - https://decrypt.co/358129/openclaw-creator-offers-acquire-ai-sensation-stay-open-source
