The AI Lab
Long-form essays, analysis, and founder commentary on AI content provenance, verified human identity, and the institutional infrastructure of the post-AI internet.
Institutional Substack
The AI Lab on Substack
Editorial notes on TIP, the AI Trust Council, standards comparisons, regulatory crosswalks. Roughly weekly.
Subscribe at theailaborg.substack.com
Founder Substack
Dinesh Mendhe on Substack
First-person founder essays. Industry observations, predictions outside the Council voice. Roughly twice a month.
Subscribe at dineshmendhe.substack.com
Featured
Policy Analysis
Shadow AI now shows up in 43% of security incidents, roughly double a year ago, while more than two thirds of organizations still have no way to detect it. You cannot police every browser tab. You can verify what comes out the other end.
More from the lab
Policy Analysis
EU Policy and Regulatory Affairs Lead Anuraag Karangle on why Europe has to move from compliance claims to cryptographic proof. Article 50 of the EU AI Act became enforceable on August 2, 2026, but the operational reality is failing: watermarks are cropped, metadata is stripped on upload, automated fact-checking tools have confirmed synthetic media as authentic, and deepfake fraud targeting European enterprises and public figures continues to scale. Reactive detection is a losing race, because the cost of producing a convincing fake falls every year while the cost of inspecting everything rises. Truth as a Service (TaaS) inverts the question: instead of asking whether a file is synthetic after the fact, it embeds verifiable provenance at the point of creation. The Trust Identity Protocol (TIP) operationalizes TaaS with post-quantum ML-DSA signatures binding verified human identity and content lineage into a tamper-evident record anchored to a federated DAG (TIP Network) ledger, governed independently by the AI Trust Council so that the standard remains open, neutral, and outside the control of any single company or government.
Policy Analysis
EU Policy and Regulatory Affairs Lead Anuraag Karangle on why static AI labeling fails EU AI Act Article 50, and why real compliance is cryptographic. Watermarks can be cropped in seconds and standard metadata is stripped the moment content is uploaded to a social platform or sent through a messaging app, so a self-declared badge is a liability, not compliance. The Trust Identity Protocol (TIP) signs AI content and chatbot interactions at the moment of creation with post-quantum ML-DSA signatures anchored to a decentralized, federated DAG (TIP Network) ledger, so the proof of origin survives cropping, copying, and metadata stripping and stays verifiable in real time. The independent AI Trust Council governs how identities are verified and how AI alignment is certified. The AI Lab is inviting enterprises, technology providers, and academic institutions to join as Node Partners and Verification Providers before the August 2, 2026 deadline.
Leadership Essay
Why The AI Lab built the Trust Identity Protocol: an open, independently governed standard for verified human identity, content provenance, and AI trust.
Council Commentary
AI Trust Council Independent Founding Member Joshua D. Baron published 'Trust Cannot Be Engineered Behind Closed Doors' on LinkedIn. His essay frames why trust in AI-era content requires open standards, multi-stakeholder governance, and a Charter that no single company can capture. The argument is the operating principle of the Trust Identity Protocol (TIP) and the AI Trust Council convened May 3, 2026.
Publication
The Trust Identity Protocol Whitepaper Version 1.0 is the canonical public specification of the open standard that lets any real person prove they are human, and any piece of content declare how it was made. The whitepaper documents the three-layer architecture (TIP-ID, TIP-CONTENT, TIP-TRUST), the post-quantum cryptographic foundation (ML-DSA-65, ML-KEM-768, SLH-DSA), the federated DAG (TIP Network), the AI Trust Council governance posture under Article 95(3) of the EU AI Act, the regulatory alignment across the United States, United Kingdom, European Union, New Zealand, and India, the TIPCL-1.0 licensing schedule with conversion to Apache 2.0 in 2031, the universal seventeen-type content taxonomy, the verified endpoint inventory matching the canonical specification v5.0, the Trust Score tier system (HIGHLY_TRUSTED, TRUSTED, REVIEW_ADVISED, LOW_TRUST, NOT_TRUSTED), and the public reception of the protocol to date.
Compliance Playbook
Comprehensive operational playbook for EU AI Act compliance built on the Trust Identity Protocol (TIP). Maps Article 50 paragraphs 50(1) through 50(5) directly to the four TIP Origin Codes (OH, AA, AG, MX). Explains why detection-based compliance fails and declaration-based compliance scales. Documents the AI Trust Council as the codes-of-practice body Article 50(6) and 50(7) explicitly anticipate. Includes the 30-day compliance sprint, the cost of non-compliance (up to fifteen million euros or three percent of global turnover under Article 99), the free-tier eligibility for journalism, nonprofits, governments, education, and small businesses under one hundred thousand dollars revenue, and a cross-jurisdictional crosswalk covering the US NIST AI RMF, UK AI Safety Institute, India DPDP Act, Australia AI Ethics Principles, and Brazil PL 2338. The complete AI-powered compliance reference for any organization operating AI systems that touch European users.
Launch Announcement
The formal public launch announcement of the Trust Identity Protocol (TIP). Explains the three layers (TIP-ID, TIP-CONTENT, TIP-TRUST), the four Origin Codes (OH, AA, AG, MX), the post-quantum cryptographic foundation (FIPS 203, 204, 205), how TIP compares to C2PA and JPEG Trust, the independent AI Trust Council governance, the TIPCL-1.0 licensing schedule with conversion to Apache 2.0 in 2031, current production traction (two million registrations, eleven institutional partners, four government engagements), and the operational integration paths for creators, publishers, developers, and Verification Providers.
Founder Essay
An essay on why the period 2026 to 2036 will become known as the Authenticity Decade, the structural parallel between HTTPS adoption and verified-origin adoption, the three-body institutional problem of internet trust, and five concrete predictions for how content provenance becomes routine internet infrastructure.
Founder Essay
An honest, side-by-side reading of the Coalition for Content Provenance and Authenticity (C2PA) standard and the Trust Identity Protocol (TIP). The essay steelmans C2PA's three substantial wins, then names the three gaps that motivated TIP: identity, federation, and post-quantum cryptography. Closes with where the two standards should converge, and how publishers, regulators, and platforms should think about both.
Founder Essay
An announcement and short essay on why The AI Lab publishes its editorial content via RSS and JSON Feed. Connects the decision to the Trust Identity Protocol's open-standards thesis, explains how to subscribe in any major reader, identifies the practical operational paths for journalists, regulators, and researchers, and previews the TIP-signed RSS items planned for the period after the Genesis Date.
Founder Essay
A founder essay in nine sections. The catalyzing morning. The technical gap C2PA was never going to fix. The institutional gap consortia cannot fill. The choice of Delaware. The reasoning behind the AI Trust Council. Five things I got wrong in 2024. Three things I am afraid of now. Five predictions you can hold me to. A direct address to journalists, regulators, high-tech CEOs, and individuals who want to be part of the work.