Grok Bot

Original Article ↗ Hacker News Discussion ↗

Grok Bot, xAI’s new always-on agent that runs in its own cloud VM and can log into users’ apps to act on their behalf, is drawing equal parts fascination and alarm. Commenters see clear productivity potential in delegating tasks like sourcing suppliers or buying tickets, but raise serious concerns about handing vast account access and sensitive data to an Elon Musk–controlled platform, as well as unresolved issues like prompt injection, bot detection, and runaway token costs. Many expect this “AI employee” model to spread, yet argue that trust, security, legal liability, and open, provider-agnostic alternatives will ultimately determine whether such tools become viable or dangerous infrastructure.

Product & Pricing

  • Grok Bot is positioned as an always-on “agent OS” tied to high‑tier Grok/Cursor plans (~$120–$200/user/month).
  • Bots run in their own Linux VMs, can use a browser, and persist state, routines, and “skills” per domain.
  • Some see this as the natural evolution from autocomplete → chat → agents, and a clever product move; others see it as yet another similar in‑house agent system.

Trust, Brand, and Data Concerns

  • Many commenters say they will not trust anything from X/xAI with their data, citing prior behavior, rebranding choices, and recent radical political rhetoric from the owner.
  • There is concern that Cursor’s acquisition “taints” that brand as a wedge into enterprises.
  • Some consider state‑controlled foreign platforms less worrying than billionaire‑controlled ones; others see both as bad in different ways.

Security, Credentials & Liability

  • Core worry: bots running 24/7 with access to email, files, bank or SaaS accounts.
  • The demo shows the bot taking over browser sessions/logins; people fear credential capture, session hijacking, and that users, not vendors, will bear legal/accountability risk.
  • Prompt injection is viewed as fundamentally unsolved; quoted claims that it is “largely solved” are heavily disputed due to low but non‑zero failure rates and evolving attack surfaces.
  • Some argue bots should have separate, least‑privilege accounts and payment methods; others note current SaaS per‑seat pricing and weak permission models make this expensive and awkward.

Technical Merits & Costs

  • Positive reports: always‑on agents that can negotiate with vendors, manage tickets, or triage bugs feel powerful and natural; domain‑separated bots that talk to each other can work well.
  • Major downside: enormous token usage; persistent agents can exceed years of prior LLM usage in a month. Seen as only viable if tokens get vastly cheaper.
  • Skepticism that computer use is “solved”; existing tools often break on bot blockers, CAPTCHAs, and login‑walled sites. How Grok Bot really handles this at scale is unclear.

Ecosystem, Alternatives & Social Impact

  • Comparisons to OpenClaw, Hermes, Claude Code, and self‑hosted/open‑source agent frameworks; some argue vendor‑lock‑in is needless given open models.
  • Widespread fear of “agent spam”: RFQs, recruiting, email, and content all generated at scale by bots, forcing recipients to deploy their own filters and agents.
  • Several see this as deeply anti‑human and economically wasteful; others argue it’s just the next phase of automation and that AI‑mediated interactions may become the default.