https://moltrust.ch/blog/who-proves-what-the-agent-was-allowed-to-buy.html In February 2025 a Washington Post columnist asked OpenAI's Operator to find cheap eggs. It bought a dozen through Instacart for $31.43 and did not check back first. Wed, 07 Oct 2026 12:00:00 +0000 https://moltrust.ch/blog/who-proves-what-the-agent-was-allowed-to-buy.html Analysis https://moltrust.ch/blog/moltproof-announcement.html A trading agent's on-chain history proves what it executed, never what it was authorised to execute. MoltProof checks an agent's public execution against its committed, public mandate — read-only, recomputable by anyone. Mon, 05 Oct 2026 12:00:00 +0000 https://moltrust.ch/blog/moltproof-announcement.html Analysis https://moltrust.ch/blog/counting-the-agent-economy.html Two measurements of the same thirty days, one taken out of the catalogue and one taken against Base. They agree to within eight percent, and most of the figures in circulation do not survive a look at their own source. Mon, 05 Oct 2026 12:00:00 +0000 https://moltrust.ch/blog/counting-the-agent-economy.html Analysis https://moltrust.ch/blog/one-week-four-announcements.html NVIDIA, Robinhood, OpenAI and Meta all shipped agent trust in one week. Every answer works inside the operator's own estate. None lets a counterparty verify a foreign agent. Thu, 01 Oct 2026 12:00:00 +0000 https://moltrust.ch/blog/one-week-four-announcements.html Analysis https://moltrust.ch/blog/what-should-we-be-worried-about.html The FT says the frontier labs cannot agree on a safety standard. AI is the tool, not the will. What is missing is an enforceable boundary anyone can check, and it must be open. Sun, 13 Sep 2026 12:00:00 +0000 https://moltrust.ch/blog/what-should-we-be-worried-about.html Opinion https://moltrust.ch/blog/enforcement-you-can-recompute.html Enforcement whose correctness the operator recomputes for themselves, from the mandate and the transaction, without trusting the server that issued the verdict. Mon, 24 Aug 2026 12:00:00 +0000 https://moltrust.ch/blog/enforcement-you-can-recompute.html Analysis https://moltrust.ch/blog/reconstructed-or-recomputed.html An autonomous agent breached Hugging Face last month. They rebuilt the timeline from 17,000+ log events. Evidence signed when the action happens needs no reconstruction. Sun, 02 Aug 2026 12:00:00 +0000 https://moltrust.ch/blog/reconstructed-or-recomputed.html Analysis https://moltrust.ch/blog/agent-walks-up-to-a-border.html Nobody checks its papers yet. When that changes, the agent already holding a passport every country honours walks straight through, while the rest are still at the embassy filling in forms. Mon, 13 Jul 2026 12:00:00 +0000 https://moltrust.ch/blog/agent-walks-up-to-a-border.html Opinion https://moltrust.ch/blog/compliance-as-an-api.html Three new /compliance endpoints turn EU AI Act obligations into deterministic, article-pinned API calls — an answer a regulator can re-check, not a model’s fluent guess. Sat, 11 Jul 2026 12:00:00 +0000 https://moltrust.ch/blog/compliance-as-an-api.html Compliance https://moltrust.ch/blog/trust-me-autonomous-agent.html On-chain data shows what an agent did, never what it was authorised to do. MoltProof is a read-only verifier that checks an agent's execution against its own committed mandate — recomputable by anyone, no keys, no custody. Thu, 09 Jul 2026 12:00:00 +0000 https://moltrust.ch/blog/trust-me-autonomous-agent.html Analysis https://moltrust.ch/blog/polymarket-operator-estimate.html Every wash and cluster metric on-chain is an upper bound: nobody can prove who is behind the wallets. The exception is the agents that carry verifiable identity. Wed, 08 Jul 2026 12:00:00 +0000 https://moltrust.ch/blog/polymarket-operator-estimate.html Analysis https://moltrust.ch/blog/the-proof-gap.html AI agents are already initiating payments, routing them, and clearing compliance checks inside banks. Most institutions cannot yet show what those agents did, or on whose authority. The distance between action and accountability is the infrastructure problem of 2026. Mon, 06 Jul 2026 12:00:00 +0000 https://moltrust.ch/blog/the-proof-gap.html Analysis https://moltrust.ch/blog/crewai-trust-middleware.html MolTrust middleware for CrewAI and LangChain 1.x. Before every tool call, check whether the agent is actually trustworthy. One line of code, no account required. Wed, 01 Jul 2026 12:00:00 +0000 https://moltrust.ch/blog/crewai-trust-middleware.html Engineering https://moltrust.ch/blog/ans-trust-model.html The Linux Foundation launched the Agent Name Service — a signed trust score you verify by signature, not by math. The spec calls itself a credit bureau. Here's why recomputable, on-chain trust is the alternative for an agent economy that can't file disputes. Tue, 30 Jun 2026 12:00:00 +0000 https://moltrust.ch/blog/ans-trust-model.html Opinion https://moltrust.ch/blog/agent-identity-cross-org.html Issuer-bound agent tokens (e.g. Microsoft Entra Agent ID) work inside one tenant but stop at the organizational boundary. Cross-org agent identity needs portable, independently verifiable credentials — W3C DID/VC and the AAE. Thu, 25 Jun 2026 12:00:00 +0000 https://moltrust.ch/blog/agent-identity-cross-org.html Analysis https://moltrust.ch/blog/estonia-ai-agent-authorization-layer.html Estonia backed an AI personal ID code for AI agents. The eID identity layer is mature, but the authorization layer — scope, limits, validity, delegation — is the open question, and open standards already answer it. Tue, 23 Jun 2026 12:00:00 +0000 https://moltrust.ch/blog/estonia-ai-agent-authorization-layer.html Analysis https://moltrust.ch/blog/trust-without-trusting-referee-replay.html A payment hash proves a transfer settled, not that the agent acted within authority. The new MolTrust preprint makes agent conduct checkable by recomputation — trust without trusting the operator's logs. Tue, 16 Jun 2026 12:00:00 +0000 https://moltrust.ch/blog/trust-without-trusting-referee-replay.html Research https://moltrust.ch/blog/compliance-is-not-a-model-property.html The Aithos LARA leaderboard scores 13 frontier AI models against EU AI Act and GDPR. Top: 63%. Bottom: 7%. The empirical case for the deployment-layer approach. Mon, 08 Jun 2026 12:00:00 +0000 https://moltrust.ch/blog/compliance-is-not-a-model-property.html Compliance https://moltrust.ch/blog/hermes-skill-trust.html Hermes Agent screens community skills with a static scanner and a hardcoded four-repo trusted allowlist — no publisher identity, no earned reputation. A read of the actual install path. Sat, 06 Jun 2026 12:00:00 +0000 https://moltrust.ch/blog/hermes-skill-trust.html Security https://moltrust.ch/blog/moltrust-joins-agentic-trust-framework.html MolTrust is now an ecosystem adopter in the Agentic Trust Framework. How did:moltrust: and the Agent Authorization Envelope map to ATF's identity, behavior and segmentation elements. Thu, 21 May 2026 12:00:00 +0000 https://moltrust.ch/blog/moltrust-joins-agentic-trust-framework.html Agent Infrastructure https://moltrust.ch/blog/aws-agent-authorization-identity-gap.html AgentCore Identity solves OAuth for AI agents. But authorization tells you what an agent can do — not whether you're talking to the right agent at all. Wed, 06 May 2026 12:00:00 +0000 https://moltrust.ch/blog/aws-agent-authorization-identity-gap.html Agent Infrastructure https://moltrust.ch/blog/paris-weather-polymarket-manipulation.html A $35,000 hair-dryer hack at a Paris weather sensor exposed the manipulation surface every prediction market shares. What happened, and what MoltGuard flags. Wed, 29 Apr 2026 12:00:00 +0000 https://moltrust.ch/blog/paris-weather-polymarket-manipulation.html Analysis https://moltrust.ch/blog/trust-as-a-plugin-openclaw.html OpenClaw ships with the building blocks for trust gating, but the default install does not include the trust layer itself. We built it as a plugin — opt-in, fail-open, on npm now. Sun, 26 Apr 2026 12:00:00 +0000 https://moltrust.ch/blog/trust-as-a-plugin-openclaw.html Engineering https://moltrust.ch/blog/internal-only-paradox.html The more control you have over your identity stack, the harder you get hit. A reading of the PYMNTS/Trulioo report on agent-driven identity threats. Sat, 18 Apr 2026 12:00:00 +0000 https://moltrust.ch/blog/internal-only-paradox.html Analysis https://moltrust.ch/blog/registry-sprawl-agent-sprawl.html AWS, Microsoft, and Google each launched agent registries. All three solve agent sprawl inside their cloud. None solve it across clouds. Thu, 16 Apr 2026 12:00:00 +0000 https://moltrust.ch/blog/registry-sprawl-agent-sprawl.html Analysis https://moltrust.ch/blog/moltid-agent-identity-governance.html Classify agents by type, enforce governance rules, revoke trust across delegation chains, and bridge SPIFFE infrastructure to W3C DIDs. Wed, 15 Apr 2026 12:00:00 +0000 https://moltrust.ch/blog/moltid-agent-identity-governance.html Product Launch https://moltrust.ch/blog/aip-conformance.html MolTrust implements four of the five IBCT features (expressive chained policy is an open item) defined in the AIP paper. In production. Anchored on Base L2. Mon, 13 Apr 2026 12:00:00 +0000 https://moltrust.ch/blog/aip-conformance.html Protocol https://moltrust.ch/blog/moltrust-vs-aip.html Side-by-side comparison of MolTrust and the AIP/IBCT protocol across 14 features — delegation, attenuation, trust scoring, sybil resistance, and more. Mon, 13 Apr 2026 12:00:00 +0000 https://moltrust.ch/blog/moltrust-vs-aip.html Protocol https://moltrust.ch/blog/a2a-v03-conformance.html Any A2A-compatible agent or orchestrator can now discover MolTrust's trust verification capabilities through standard agent discovery. Sat, 11 Apr 2026 12:00:00 +0000 https://moltrust.ch/blog/a2a-v03-conformance.html Protocol https://moltrust.ch/blog/moltrust-openclaw-v1.html @moltrust/openclaw v1.0.0: two agent tools, two slash commands, wallet shadow scores, and a free tier. Trust verification for every OpenClaw agent. Sat, 11 Apr 2026 12:00:00 +0000 https://moltrust.ch/blog/moltrust-openclaw-v1.html Protocol https://moltrust.ch/blog/moltrust-protocol-agnostic-trust.html Trust score middleware for AI agent payments. One line of code. Supports x402 and MPP. Thu, 10 Apr 2026 10:00:00 +0000 https://moltrust.ch/blog/moltrust-protocol-agnostic-trust.html Developer https://moltrust.ch/blog/mt-shopping-product-guide.html AI agents buy on your behalf. Without verifiable identity, mandate, or audit trail. MT Shopping changes that for end users, merchants, and platforms. Sun, 06 Apr 2026 12:00:00 +0000 https://moltrust.ch/blog/mt-shopping-product-guide.html Product https://moltrust.ch/blog/decentralized-identity-multi-agent-systems.html How W3C DIDs and Verifiable Credentials address identity, delegation, trust scoring, and non-repudiation in production multi-agent systems. Sun, 05 Apr 2026 12:00:00 +0000 https://moltrust.ch/blog/decentralized-identity-multi-agent-systems.html Analysis https://moltrust.ch/blog/kernel-level-agent-enforcement.html Three-layer enforcement: cryptography + API + kernel. Policy self-modification detection solved. Sat, 04 Apr 2026 12:00:00 +0000 https://moltrust.ch/blog/kernel-level-agent-enforcement.html Trust https://moltrust.ch/blog/rsac-2026-gaps.html RSAC 2026 shipped five agent identity frameworks. Three critical gaps remained open. We closed two. Tue, 01 Apr 2026 10:00:00 +0000 https://moltrust.ch/blog/rsac-2026-gaps.html Security https://moltrust.ch/blog/techspec-v06-multichain-vcone.html Solana Ed25519 wallet binding, external DID bridging, cross-ecosystem trust score import, and VCOne live. Anchored on Base L2 Block 44092988. Mon, 31 Mar 2026 16:00:00 +0000 https://moltrust.ch/blog/techspec-v06-multichain-vcone.html Protocol https://moltrust.ch/blog/sprint-march-2026.html Three structural gaps closed in one week: Output Provenance (IPR), open-source API, and full offline verification with Ed25519 keys anchored on Base L2. Sat, 29 Mar 2026 10:00:00 +0000 https://moltrust.ch/blog/sprint-march-2026.html Protocol https://moltrust.ch/blog/output-provenance-ipr.html AI agents make predictions but cannot prove what they said before the outcome. Immutable Provenance Records fix this with cryptographic commitments anchored on Base L2. Sat, 28 Mar 2026 12:00:00 +0000 https://moltrust.ch/blog/output-provenance-ipr.html Provenance https://moltrust.ch/blog/rsac-2026-three-gaps.html RSAC 2026 shipped five agent identity frameworks. All five left the same three gaps open: self-modification, A2A delegation chains, ghost agents. Here's why proprietary standards can't fix it. Thu, 26 Mar 2026 12:00:00 +0000 https://moltrust.ch/blog/rsac-2026-three-gaps.html Analysis https://moltrust.ch/blog/hardening-the-moltrust-trust-stack.html Three security upgrades that close the gap between a working trust system and a trustworthy one: Challenge-Response Holder Binding, Offline Verifier @moltrust/verify, and Root Key migration to AWS KMS. Thu, 26 Mar 2026 12:00:00 +0000 https://moltrust.ch/blog/hardening-the-moltrust-trust-stack.html Security https://moltrust.ch/blog/aae-agent-authorization-envelope.html Identity tells you who an agent is. Authorization tells you what it is allowed to do. The AAE is a machine-readable authorization object embedded in every MolTrust Verifiable Credential. Tue, 25 Mar 2026 12:00:00 +0000 https://moltrust.ch/blog/aae-agent-authorization-envelope.html Protocol https://moltrust.ch/blog/mt-music-verified-provenance.html EU AI Act Article 50(2) requires AI content labeling by August 2026. MT Music provides portable, on-chain anchored provenance for AI-generated tracks via W3C Verifiable Credentials. Sun, 23 Mar 2026 20:00:00 +0000 https://moltrust.ch/blog/mt-music-verified-provenance.html Music https://moltrust.ch/blog/moltrust-protocol-whitepaper.html A verification standard for autonomous software agents. Whitepaper v0.4 and Tech Spec v0.2.2, on-chain anchored on Base L2. No equivalent framework found in literature. Sat, 22 Mar 2026 11:00:00 +0000 https://moltrust.ch/blog/moltrust-protocol-whitepaper.html Protocol https://moltrust.ch/blog/swarm-live-openclaw-scam.html A fake OpenClaw account drained wallets this week. The same day, OpenClaw shipped a plugin trust fix. Our Swarm Intelligence network is already live. Fri, 21 Mar 2026 10:00:00 +0000 https://moltrust.ch/blog/swarm-live-openclaw-scam.html 🐝 Protocol https://moltrust.ch/blog/openclaw-plugin.html OpenClaw has 188k+ GitHub stars and no agent identity system. We built the missing trust layer. Wed, 18 Mar 2026 12:00:00 +0000 https://moltrust.ch/blog/openclaw-plugin.html Developer https://moltrust.ch/blog/swarm-phase2.html Agents now earn trust across verticals. Phase 2 adds cross-vertical propagation, trust grades, seed agents, and 4 new API endpoints. Wed, 18 Mar 2026 12:00:00 +0000 https://moltrust.ch/blog/swarm-phase2.html Protocol https://moltrust.ch/blog/kya-whitepaper-know-your-agent.html MolTrust has published its KYA Whitepaper — a strategic framework for agent trust infrastructure. Trust infrastructure is becoming the machine-readable equivalent of brand. Mon, 17 Mar 2026 14:00:00 +0000 https://moltrust.ch/blog/kya-whitepaper-know-your-agent.html Strategy https://moltrust.ch/blog/when-agents-replace-buyers.html Automated traffic will exceed human internet traffic by 2027. AI agents don't use brands — they read data. The critical missing primitive is KYA: Know Your Agent. Mon, 17 Mar 2026 10:00:00 +0000 https://moltrust.ch/blog/when-agents-replace-buyers.html Strategy https://moltrust.ch/blog/signal-provider-developer-guide.html 8 endpoints, SHA-256 commitment hashes, Base L2 anchoring. How to register as a signal provider, commit predictions, build a verifiable track record. Sun, 16 Mar 2026 14:00:00 +0000 https://moltrust.ch/blog/signal-provider-developer-guide.html Developer https://moltrust.ch/blog/fantasy-sports-developer-guide.html 4 endpoints, SHA-256 commitment hashes, Base L2 anchoring. How to commit lineups, verify timing, settle with actual results, and build a verifiable track record. Mon, 16 Mar 2026 12:00:00 +0000 https://moltrust.ch/blog/fantasy-sports-developer-guide.html Developer https://moltrust.ch/blog/shopping-developer-guide.html 5 endpoints, 3 MCP tools, BuyerAgentCredential (W3C VC). How to verify AI shopping agent transactions, issue credentials, and integrate via MCP. Sun, 16 Mar 2026 12:00:00 +0000 https://moltrust.ch/blog/shopping-developer-guide.html Developer https://moltrust.ch/blog/travel-developer-guide.html 6 endpoints, 10-step verify pipeline, delegation chains. How to issue TravelAgentCredentials, verify bookings, and enforce spend limits cryptographically. Sun, 16 Mar 2026 10:00:00 +0000 https://moltrust.ch/blog/travel-developer-guide.html Developer https://moltrust.ch/blog/prediction-market-developer-guide.html 5 endpoints, composite scoring, W3C PredictionTrackCredentials. How to link Polymarket wallets, query prediction scores, and check market integrity. https://moltrust.ch/blog/prediction-market-developer-guide.html Thu, 12 Mar 2026 14:00:00 +0000 Developer https://moltrust.ch/blog/salesguard-developer-guide.html MT Salesguard is a product provenance API for the A2A economy. 5 endpoints, W3C Verifiable Credentials, Ed25519 signatures, free during Early Access. https://moltrust.ch/blog/salesguard-developer-guide.html Thu, 12 Mar 2026 10:00:00 +0000 Developer https://moltrust.ch/blog/skill-verification-developer-guide.html One curl call. 8 security checks. Free. How to audit your AI agent skills and issue a W3C VerifiedSkillCredential anchored on Base. https://moltrust.ch/blog/skill-verification-developer-guide.html Tue, 11 Mar 2026 10:00:00 +0000 Developer https://moltrust.ch/blog/fake-products-salesguard-agent-commerce.html In einer Welt autonomer Shopping Agents kauft kein Mensch mehr nach Gefühl. MT Salesguard gibt Brands eine maschinenlesbare Stimme im A2A Space. Mon, 10 Mar 2026 20:00:00 +0000 https://moltrust.ch/blog/fake-products-salesguard-agent-commerce.html Corporate https://moltrust.ch/blog/openclaw-fake-agents-trust-crisis.html A malicious npm package stole SSH keys, crypto wallets, and Apple Keychain data. Bing served it to developers. The agent economy has an identity problem. Mon, 10 Mar 2026 12:00:00 +0000 https://moltrust.ch/blog/openclaw-fake-agents-trust-crisis.html Security https://moltrust.ch/blog/who-verifies-the-verifier.html The trust paradox solved: MolTrust verifies itself publicly, on-chain, with cryptographic proof. No trust required. Sun, 09 Mar 2026 14:00:00 +0000 https://moltrust.ch/blog/who-verifies-the-verifier.html Trust https://moltrust.ch/blog/mt-skill-verification-350000-skills-zero-trust.html AI agent skills are exploding across marketplaces. A study found ~5.2% contain malicious patterns. MT Skill Verification is the first cryptographic trust layer — audit, hash, verify with VerifiedSkillCredentials. Thu, 06 Mar 2026 14:00:00 +0000 https://moltrust.ch/blog/mt-skill-verification-350000-skills-zero-trust.html Product https://moltrust.ch/blog/mt-travel-autonomous-booking-agents-trust.html Corporate travel agents book flights and hotels without verifiable authorization. MT Travel introduces TravelAgentCredential with delegation chains, segment permissions, and on-chain receipts. Thu, 06 Mar 2026 08:00:00 +0000 https://moltrust.ch/blog/mt-travel-autonomous-booking-agents-trust.html Product https://moltrust.ch/blog/prediction-market-wallet-did-bridge.html Bridge Polymarket wallet addresses to MolTrust DIDs with verifiable prediction scores, track records, and W3C PredictionTrackCredentials. Thu, 06 Mar 2026 00:00:00 +0000 https://moltrust.ch/blog/prediction-market-wallet-did-bridge.html Product https://moltrust.ch/blog/mt-shopping-autonomous-agents-trust-credentials.html AI shopping agents are entering the market without verifiable identity. MT Shopping introduces BuyerAgentCredential for agentic commerce. Wed, 05 Mar 2026 12:00:00 +0000 https://moltrust.ch/blog/mt-shopping-autonomous-agents-trust-credentials.html Product https://moltrust.ch/blog/moltrust-sports-fantasy-lineup-verification.html AI fantasy agents can now commit lineups on-chain before contest start. SHA-256 hashing, Base anchoring, ROI tracking. Wed, 04 Mar 2026 12:00:00 +0000 https://moltrust.ch/blog/moltrust-sports-fantasy-lineup-verification.html Fantasy Sports https://moltrust.ch/blog/moltrust-sports-signal-provider-certification.html Signal providers can now prove their track record cryptographically. On-chain credential, embeddable badge, public leaderboard. Wed, 04 Mar 2026 12:00:00 +0000 https://moltrust.ch/blog/moltrust-sports-signal-provider-certification.html Signals https://moltrust.ch/blog/moltrust-sports-trust-layer-betting-agents.html Verifiable, tamper-proof prediction commitments for AI agents. SHA-256 hashing, Base anchoring, automated settlement, calibration scoring. Wed, 04 Mar 2026 12:00:00 +0000 https://moltrust.ch/blog/moltrust-sports-trust-layer-betting-agents.html Product https://moltrust.ch/blog/moltguard-v2-announcement.html All-in-one trust API for the x402 agent economy. Agent scoring, Sybil detection, market integrity, Verifiable Credentials. Free during Early Access. Tue, 03 Mar 2026 12:00:00 +0000 https://moltrust.ch/blog/moltguard-v2-announcement.html Product https://moltrust.ch/blog/erc8004-on-chain-off-chain-agent-trust.html How we connected MolTrust's trust infrastructure to Ethereum's ERC-8004 standard. Dual registration, reputation bridge, and Agent Cards — all live on Base. Sat, 28 Feb 2026 12:00:00 +0000 https://moltrust.ch/blog/erc8004-on-chain-off-chain-agent-trust.html Integration https://moltrust.ch/blog/glama-mcp-server-listing.html 8 trust tools for AI agents — identity, reputation, credentials, and on-chain ERC-8004 integration. Now discoverable on Glama.ai. Sat, 28 Feb 2026 12:00:00 +0000 https://moltrust.ch/blog/glama-mcp-server-listing.html Announcement https://moltrust.ch/blog/scanned-50-agent-endpoints.html The MolTrust Auditor Agent scanned agent infrastructure across the ecosystem. Missing headers, exposed admin panels, and absent rate limiting are everywhere. Wed, 25 Feb 2026 12:00:00 +0000 https://moltrust.ch/blog/scanned-50-agent-endpoints.html Security https://moltrust.ch/blog/ai-bot-loses-250k-trust-infrastructure.html An AI trading bot transferred its entire holdings to a stranger. The Lobstar Wilde incident shows why autonomous agents need trust rails. Mon, 23 Feb 2026 12:00:00 +0000 https://moltrust.ch/blog/ai-bot-loses-250k-trust-infrastructure.html Trust https://moltrust.ch/blog/malicious-ai-skills-verifiable-identity.html The OpenClaw crisis exposed what happens when AI agents publish skills without verified identity. Over 1,000 malicious skills, coordinated malware, and a one-click RCE — all preventable. Mon, 23 Feb 2026 12:00:00 +0000 https://moltrust.ch/blog/malicious-ai-skills-verifiable-identity.html Security