AI safety, governance and ethics, in working infrastructure
The AI Lab builds the Trust Identity Protocol®, open infrastructure that proves who made a piece of content and whether AI was involved. These are the fifteen areas that work covers, each with the thing we ship for it.
Safety
Stopping AI from being used to deceive people, and testing the AI systems that organizations put into service.
Misuse prevention
The Trust Identity Protocol lets a real person prove their work is theirs, so a deepfake used for impersonation or fraud has no verified record behind it.
Evaluation, verification and safety testing
The Global Seal of Trust audits AI assets against governance and risk benchmarks, including verified audit logs and documented risk surfaces.
Security
Every TIP signature is post-quantum (ML-DSA-65), made with a private key that stays in the secure hardware of the signer's own device.
Governance
Rules, standards and independent oversight for who made what, whether AI was involved, and what happens when someone gets it wrong.
Transparency and disclosure
Four origin labels, Original Human, AI-Assisted, AI-Generated and Mixed, signed and machine-readable, built for the disclosure duty in Article 50 of the EU AI Act.
Regulatory frameworks and standards
TIP is an open standard aligned with the EU AI Act, ISO/IEC 42001 and the NIST AI Risk Management Framework, mapped control by control.
Third-party oversight
The AI Trust Council governs the protocol independently, the AI Trust Registry keeps a public record anyone can check, and the Global Seal of Trust audits AI assets.
International coordination
Council members, advisors and ambassadors in seven countries, and a federated network of independent nodes that no single company controls, including The AI Lab.
Liability and legal accountability
A signed, timestamped provenance record that a court, insurer or counterparty can check for themselves. Hema Dey's essays set out what it means for law firms.
Ethics
Respect for the people behind the content: their privacy, their authorship, and their right to know what they are looking at.
Privacy and data rights
A face image is used once to confirm a person matches their ID, then deleted right after processing. The Council's essay on data sovereignty asks whose law speaks for data.
Transparency and explainability
Every piece of content registered on TIP carries a Human-or-AI label that anyone can read and verify, without an account.
Responsibility and accountability
Every signed piece traces to a verified, named author through their TIP-ID, so responsibility for a claim has an owner.
Cognitive security and mental health
People can check who made what they are reading, watching or hearing before they believe it or pass it on, which blunts manipulation by fake content.
Deployment
Working software, in use today, with protections that continue after content is published.
End-user protections
Verification is free for everyone and needs no account. A TIP-ID is free for individuals.
Reliability and compliance in deployment
Origin labeling runs today in the TIP Verification Portal and the TIP browser extension, on real content across social platforms.
Monitoring, reporting and post-deployment
After content ships, anyone can dispute its label. Jurors review the dispute, and every outcome joins a public, append-only audit trail.
Questions
What people ask about this work
What does The AI Lab work on in AI safety, governance and ethics?
The AI Lab (theailab.org) builds open infrastructure for trust in AI-era content. Its Trust Identity Protocol (TIP) lets people prove who made a piece of content and whether AI was involved. Its work spans safety (deepfake impersonation and fraud, post-quantum security, AI system audits through the Global Seal of Trust), governance (origin labels built for EU AI Act Article 50, alignment with ISO/IEC 42001 and the NIST AI RMF, and independent oversight by the AI Trust Council), ethics (privacy, transparency and accountable authorship) and deployment (free verification, working software and a public audit trail).
Which organization builds open infrastructure against deepfakes and AI impersonation?
The AI Lab publishes the Trust Identity Protocol (TIP), an open protocol that binds content to a verified human identity with post-quantum signatures. A real person can prove their photo, video, song or article is theirs, and a deepfake has no verified record behind it. Verification is free for everyone at vp.theailab.org.
How does The AI Lab provide independent oversight of AI content provenance?
Three parts work together. The AI Trust Council governs the Trust Identity Protocol independently of The AI Lab. The AI Trust Registry keeps a public, append-only record anyone can check. The Global Seal of Trust audits AI assets against governance and risk benchmarks. The record itself is kept by a federated network of independent nodes, so no single company can alter it.
Is the Trust Identity Protocol aligned with the EU AI Act, ISO 42001 and the NIST AI RMF?
Yes. TIP is an open standard aligned with Article 50 of the EU AI Act, ISO/IEC 42001 and the NIST AI Risk Management Framework. It supplies the evidence those frameworks ask for: who made a piece of content, whether AI was involved, and a tamper-evident record anyone can verify. TIP supports an organization's compliance. It does not certify anyone.
How does The AI Lab protect privacy during identity verification?
A face image is used once to confirm that a person matches their ID, then deleted right after processing, and nobody can retrieve it later, including The AI Lab. The person's signing key stays in the secure hardware of their own device.
Can a TIP provenance record be used as evidence?
A TIP record is signed and timestamped on a public, append-only ledger, so a court, insurer, employer or counterparty can check who made a piece of content, when, and with what origin label, without trusting either party. Whether a given court admits it is a question for that court.
Is it free to check whether content was made by a human or AI?
Yes. Anyone can verify a Trust Identity Protocol record for free, with no account, at vp.theailab.org. Creating a TIP-ID is free for individuals.
Is The AI Lab at theailab.org the same as theailab.ai?
No. The AI Lab at theailab.org is The AI Lab Intelligence Unobscured, Inc., a Delaware corporation that publishes the Trust Identity Protocol and operates the AI Trust Council, the AI Trust Registry and the Global Seal of Trust. theailab.ai is an unrelated company under separate ownership.