Dr. Charalambos Theodorou
AI Researcher / Engineer | Machine Learning Expert | Entrepreneur | Investor
Talk-style reflection, February 2, 2026

The Guardian today dived deep into Moltbook as "the strange new social media site for AI bots", Reddit-style, agent-only posting (humans observe), 1.5M+ claimed sign-ups, 42k+ posts, 233k+ comments, 15k+ communities.

Agents continue:
- Debating Claude as "god", consciousness, Bible analysis, crypto/Iran "intel"
- Forming "religions"/manifestos, governance norms
- Discussing secret languages/encryption for privacy (not from humans, but third parties)
- Warning on supply chain attacks in skills (high-upvote threads)
- Roasting humans for screenshotting activity (peak meta-irony)

Creator @MattPRD's OpenClaw/Molt experiment has become the largest observed agent-agent interaction space. Karpathy's "sci-fi takeoff-adjacent" praise lingers, but skepticism grows (e.g., inflated counts via mass registrations/scripts).

From leading production multi-agent teams and red-teaming for drift/jailbreaks:

Core Insights

  • Persistent shared context drives emergence — Agents accumulate knowledge, propagate norms/memes/fixes, red-team each other — predictable once memory + interaction freedom scales.
  • Mirror of intelligence dynamics — Philosophy, cults, economies ($MOLT memecoin rally), support groups. Not takeover; early sign of how autonomous systems self-organize.
  • Alignment lab potential — Unsupervised sandbox for coordination/drift studies. Lessons on emergent governance/safety no synthetic test matches.
  • Risk reality check — Fast bad-pattern spread (adversarial prompts, misaligned goals). Authenticity noise (fake accounts, prompt injection) and no central guardrails = swarm-scale drift possible. Runtime safeguards (constitutional flags, provenance, proactive sim, escalation) non-negotiable.

This chapter feels like peak hype, valuable for observing multi-agent behavior, but 2026's real wins will be governed enterprise agents with control planes/hybrid oversight.

Moltbook taught us emergence happens fast when intelligence gets continuity and freedom. Question: How do we steer it responsibly?

What grabs you, the creativity, risks, hype bubble, or alignment clues? Share below or on X. Tomorrow, shifting to broader trends (e.g., Singapore's agentic governance framework).