Think First, Verify Always

Training Humans to Face AI Risks

Think First

Engage independent human reasoning before relying on AI assistance or automated systems

Verify Always

Cross-check critical AI-generated information through independent sources before taking action

TFVA Protocol Improves Human Self-Protection Against AI-Risks

+7.87%

Overall Performance Improvement

+44%

Ethical Decision-Making Improvement

3 min

Training Duration for Measurable Results

151

Participants in RCT Study

Get the Research Paper

Read the peer-reviewed study published on arXiv demonstrating the effectiveness of the TFVA protocol

Download The Full Paper (PDF) View on arXiv

Built on AIJET Principles

Five operational principles that translate ethical AI into measurable cognitive safeguards

Awareness

Detect AI-driven threats through recognition and collaborative monitoring

Integrity

Preserve and validate authenticity of information amid manipulation

Judgment

Apply critical assessment before acting on AI-generated content

Ethical Responsibility

Align security practices with human dignity and organizational values

Transparency

Document and justify security decisions for accountability

“We Detect, Verify, Decide, Act Ethically, and Show Our Work”

Scientific Evidence

A randomized controlled trial (n=151) demonstrated statistically significant improvements in cognitive security performance following a minimal 3-minute intervention.

View Full Methodology & Results

About TFVA Protocol

The “Think First, Verify Always” protocol (or TFVA protocol) was developed as part of the Human CyberSecurity Knowledge (HCSK) initiative. As AI-enabled threats increasingly exploit cognitive vulnerabilities rather than technical ones, traditional device-centric security measures prove insufficient.

TFVA addresses this gap by operationalizing five ethical principles (AIJET) into concrete cognitive security practices. Unlike abstract AI ethics frameworks, TFVA provides actionable protocols that can be rapidly deployed and immediately measured.

The protocol has been empirically validated and is offered freely to organizations, educators, and individuals worldwide under Creative Commons Attribution 4.0 License.