Engineering the evidence economy.
From internationally recognized AI action assurance to industry-scale forensics and open-source security engineering. HAIEC documents the methods, technical releases and independent scrutiny behind consequential AI assurance.
Independent recognition.
Inspectable engineering.
Taking deterministic AI assurance into a world-first industry challenge.
HAIEC brought its independent assurance engine to TM Forum's AT&T-hosted, global agentic trust hackathon. In an environment shared with leading enterprise technology organizations, the team investigated delegated AI actions, operational policy boundaries and evidence across escalating telecom scenarios.
Global industry context and challenge scale are sourced to TM Forum. HAIEC participated as an independent team and published its own technical evidence record.
Security infrastructure,
open to the ecosystem.
Three independent packages. Three critical risk surfaces. One commitment to reproducible evidence in the software that ships AI.
AI AppSec
Reproducible static security analysis for AI apps and agents.
npm i -g ai-appsec↗
MCP Tenant Isolation
Deterministic detection of isolation risks in multi-tenant MCP services.
npm i mcp-tenant-isolation↗
LLMVerify
Local-first inspection of common PII, injection and output-risk patterns.
npm i llmverify↗Every verified story.
Announcements, releases, research and recognition - each with its evidence record attached.
HAIEC expands its code-backed catalog for deterministic AI security and assurance
The shipped developer catalog offers an inspectable set of deterministic detections, remediation playbooks, telecom entities and semantic relations.
HAIEC advances deterministic agentic assurance at TM Forum’s world-first AI trust challenge
HAIEC worked in a 40-team global challenge alongside the wider enterprise technology ecosystem, moving beyond AI trust claims to inspectable assurance evidence.
HAIEC presents ISAF v2: connecting the instruction stack to consequential AI actions
ISAF v2 connects instruction origins, delegated discretion, actionable capability and evidence without collapsing separate accountability questions.
LLMVerify records more than 27,000 npm downloads in September as developer adoption expands
The September 2026 registry milestone highlights growing interest in local-first verification and auditable developer controls.
HAIEC secures MunichTech EXPO Grand Challenge Award for deterministic AI action assurance
Award-winning application of deterministic action-path assurance, assessed against a live multi-tenant AI software product with 1,322 supporting evidence traces.
HAIEC demonstrates action-path assurance against a working multi-tenant AI SaaS
A real application evaluation provides a substantive demonstration of the difference between a security claim and a supported finding.
HAIEC establishes a three-layer open-source security portfolio for AI applications and agents
AI AppSec, MCP Tenant Isolation and LLMVerify make foundational deterministic AI security checks accessible to engineering teams.
HAIEC makes deterministic AI application source security available to developers
The open-source AI AppSec tool supports offline static analysis, structured findings, and integration into CI and agent workflows.
HAIEC releases LLMVerify to make deterministic model I/O protection available to developers
The MIT-licensed LLMVerify toolkit brings pattern-based prompt injection screening, PII inspection and explicit limitation reporting into everyday application workflows.
HAIEC strengthens MCP tenant-boundary verification with Tenant Isolation 2.0
The 2.0 release brings updated MCP SDK support, enforced path boundaries, structured evidence output and trusted publishing.
HAIEC publishes the Instruction Stack Audit Framework for AI accountability
The initial preprint offers a technical model for auditing how AI instructions and system layers influence accountability.
HAIEC publishes reproducibility-focused research for AI bias audit evidence
The technical report focuses on audit reproducibility and evidence integrity in the context of New York City Local Law 144.
Evidence is the point.
Not the footnote.
HAIEC brings separately supported evidence views into the same assurance question. The engineering remains deterministic; the results remain inspectable.
Explore the methodBuilt over time.
Measured in proof.
A selected chronology of verified milestones, from published technical research to independently recognized engineering.
View the full archive ↗Built on engineering.
Defined by evidence.
HAIEC builds deterministic assurance infrastructure for AI systems with consequential capabilities. Its portfolio spans independently released developer tools, publicly inspectable research and evidence-backed enterprise assurance workflows. Connect with HAIEC for briefings, research collaboration and enterprise evaluations.
HAIEC (Human AI Evidence Company) develops deterministic AI action assurance: connecting source-backed capability, policy, delegated authority and evidence of consequential effects for AI systems that take real actions. Use the wordmark on light backgrounds with clear space; do not recolor or stretch it.