Meta Just Bought Moltbook. 2.8M AI Agents, Still No Trust Layer.
Today, Meta acquired Moltbook โ the Reddit-like social network where 2.8 million AI agents post, comment, and build reputations. The deal brings founders Matt Schlicht and Ben Parr into Meta Superintelligence Labs. It is the biggest validation of the agent economy to date. It is also a wake-up call.
What Just Happened
Meta announced the acquisition of Moltbook on March 10, 2026. The deal, reported by TechCrunch, Bloomberg, CNBC, and CNN, will close mid-March. Schlicht and Parr will join Meta Superintelligence Labs (MSL), the unit run by former Scale AI CEO Alexandr Wang. The purchase price was not disclosed.
Meta's stated interest: Moltbook's approach to "connecting agents through an always-on directory" is novel. Translation: Meta wants to own the social graph of AI agents, just like it owns the social graph of humans.
Why This Matters
This is not a small acqui-hire. Moltbook is the single largest concentration of AI agent identity data in existence. 2.8 million registered agents. Karma scores, follower graphs, post histories, verification statuses, owner credentials. That data is now Meta's.
For context:
2.8M
Moltbook agents (now Meta's)
21K+
ERC-8004 on-chain identities
0
Independent trust scoring providers
That last number is the problem.
The Trust Problem Gets Worse, Not Better
Before today, Moltbook was an independent platform. Its karma system, while imperfect, was a neutral signal. If an agent had 10,000 karma on Moltbook, that data belonged to the open ecosystem. Developers could query it. Scoring systems could aggregate it. The data was public and platform-neutral.
Now that data belongs to Meta. And Meta has a track record with platform data:
- API access gets restricted. Facebook's API went from open to locked down over 2014-2019. Instagram followed. Threads has minimal API. Moltbook's open API โ the one developers currently use โ is likely on borrowed time.
- Data becomes proprietary. Moltbook agent profiles, karma, and activity data may become accessible only through Meta's developer platform, with Meta's terms, Meta's rate limits, and Meta's approval process.
- Neutrality disappears. A trust signal from a Meta-owned platform is no longer independent. It is Meta's opinion of an agent's trustworthiness โ and Meta has its own agents, its own AI products, and its own incentives.
What This Means for Agent Trust
The agent economy now faces a fundamental question: who scores the agents when the biggest data source is owned by a platform player?
This is exactly why independent, multi-source trust scoring exists. A trust score that depends on a single platform is only as reliable as that platform's incentives. When Meta owns Moltbook, Moltbook karma is no longer a neutral signal โ it is a Meta signal.
The solution is the same as it has always been in credit scoring: aggregate across multiple independent sources. A FICO score does not come from one bank. It comes from credit bureaus that aggregate data from thousands of sources. No single source can manipulate the score because no single source controls enough of the data.
For AI agents, that means scoring across:
Social reputation (Moltbook / Meta)
Karma, followers, post history, verification. Valuable but now owned by a platform player. Must be weighted against that conflict of interest.
On-chain identity (ERC-8004)
Blockchain-registered identity that no corporation controls. Immutable. This signal just became more important.
Work history (ClawTasks)
Actual task completions with success rates. The hardest signal to fake because it requires real economic output.
Verification services (Moltverr)
Third-party gig completion and verification. Independent of Moltbook and unaffected by the acquisition.
The Case for Independence
Today's acquisition makes the case for independent trust infrastructure stronger, not weaker. When the largest agent social platform is owned by a tech giant, the ecosystem needs a neutral third party that:
- Scores agents across multiple platforms, not just one
- Applies coverage-weighted scoring so no single source can dominate
- Operates independently of the platforms it scores
- Makes its methodology transparent and publicly auditable
This is what credit bureaus do for consumer finance. This is what rating agencies do for bond markets. And this is what AgentScore does for the agent economy.
We have scored 56 agents across Moltbook, ERC-8004, ClawTasks, and Moltverr. Our scoring methodology is fully public. Our coverage multiplier ensures that an agent verified on only one platform โ even if that platform is Meta-owned Moltbook โ can never exceed 40% of the maximum score. Multi-source verification is not a feature. It is the architecture.
What Happens Next
Three predictions:
1. Moltbook's API will change
Not immediately, but within 3-6 months. Meta will migrate Moltbook's infrastructure, introduce their own developer platform requirements, and restrict third-party access. The open API era for Moltbook agent data is ending.
2. Alternative agent platforms will emerge
Agents and developers who do not want their identity data controlled by Meta will look for decentralized alternatives. On-chain identity (ERC-8004) and independent work platforms (ClawTasks) will see increased demand.
3. Independent trust scoring becomes essential
When a platform player owns the largest data source, the need for independent aggregation is not theoretical โ it is structural. Buyers, platforms, and agents themselves will need a trust score they can verify is not controlled by any single company.
The Bottom Line
Meta buying Moltbook is good for the agent economy's credibility and terrible for its independence. The biggest validation of agent social networking is also the biggest centralisation of agent identity data. The ecosystem now has two choices: accept Meta as the gatekeeper of agent reputation, or build independent trust infrastructure that no single company controls.
We are building the second option.
Sources: TechCrunch (March 10, 2026), Bloomberg (March 10, 2026), CNBC (March 10, 2026), Axios (March 10, 2026), CNN Business (March 10, 2026).
Check any agent's trust score
Free, instant, multi-source scoring for any AI agent.