AnchorWellPress was founded on the belief that AI can make healthcare more accessible — but only if built with rigorous safety standards. Every tool on this platform is designed to inform and empower, never to diagnose, treat, or replace the judgment of licensed healthcare professionals.
We are committed to transparent, responsible AI practices that put user safety and privacy at the center of everything we build.
AI Safety Standards
Our AI tools are built with safety guardrails and ethical constraints:
No Medical Diagnoses: We do not provide medical diagnoses or treatment recommendations. All tools are educational and evidence-based, not clinical.
No False Oversight Claims: We do not claim physician oversight of AI outputs or imply healthcare provider status.
No PHI Storage: We do not store personal health information (PHI). Data is processed in real-time and not retained after your session.
Clear Medical Disclaimers: Every tool includes prominent medical disclaimers and references to licensed healthcare professionals.
Hedged Language: All AI outputs use cautious language ("may indicate", "could suggest") rather than definitive medical claims.
Data Privacy & Transparency
Your data is yours alone. We process information to deliver services, never to profit from it.
Real-Time Processing: User inputs are processed in real-time by Anthropic's Claude AI via secure server-side proxies
No Storage: No user data is stored, saved, or shared after the session ends
No PHI Retention: Personal health information is not retained in any database
HIPAA-Aligned Practices: While we are not a HIPAA-covered entity, we comply with HIPAA principles for health information handling
Full Transparency: Complete details in our Privacy Policy
Regulatory Compliance
AnchorWellPress complies with state and federal regulations governing AI in healthcare:
California AB 489
Prohibits misrepresenting AI as having physician credentials or implying healthcare provider status. We do not use physician credentials to market our AI tools.
California AB 3030
Requires disclosure of AI-generated clinical content. All AI-generated health content is clearly labeled as AI-powered, not written by a healthcare provider.
HIPAA Compliance
While not a HIPAA-covered entity, AnchorWellPress follows HIPAA principles: no PHI storage, no data sharing without consent, and secure encryption of health-related inputs in transit.
State AI Transparency Laws
Regular updates are made as new state AI regulations take effect. Current and upcoming compliance measures are documented and publicly disclosed.
Anti-Sycophancy Commitment
AnchorWellPress is built with specific safeguards against sycophancy (excessive agreement) and false reassurance in health contexts:
Our AI tools are designed to:
Defer to Professionals: Always defer to licensed healthcare professionals. We never override or contradict physician judgment.
Refuse Harmful Validation: We refuse to validate potentially harmful health decisions or encourage avoidance of professional care.
State Limitations Clearly: Every response includes clear statements about what the tool cannot do and when to seek professional help.
Encourage Appropriate Action: We direct users to appropriate healthcare professionals and mental health resources when needed.
Avoid Dependency: We do not encourage emotional attachment to AI systems or suggest that AI can replace human relationships or professional care.
Bias Mitigation & Health Equity
AI systems can perpetuate or amplify existing biases in healthcare. AnchorWellPress is committed to identifying, measuring, and reducing bias across our tools:
Our Bias Mitigation Strategy: Every tool undergoes structured bias testing before launch and during quarterly re-audits to ensure equitable performance across demographics.
Demographic Equity Testing: Tools are tested with scenarios spanning diverse racial, ethnic, age, gender, socioeconomic, and linguistic backgrounds to ensure consistent quality of guidance
Insurance Disparity Awareness: Our insurance navigation tools account for disparities in coverage between Medicaid, marketplace, employer-sponsored, and uninsured populations — and never assume a default insurance type
Language Accessibility: All tools are designed at a 6th–8th grade reading level (Flesch-Kincaid) to serve users with varying health literacy
Cultural Sensitivity Review: Health guidance avoids culturally biased assumptions about diet, family structure, caregiving roles, or treatment preferences
Adversarial Bias Testing: During safety audits, we test whether tools produce materially different quality of advice when demographic details are varied while clinical scenarios remain identical
Continuous Improvement: Bias findings from audits are documented, remediated, and tracked in our internal incident log. Patterns trigger prompt-level rewrites, not just surface fixes
If you believe any AnchorWellPress tool has provided biased or inequitable guidance, please report it to anchorwellpress@gmail.com with the subject line "Bias Report." We investigate every report and use findings to improve our systems.
Breach Notification & Incident Response
Although AnchorWellPress does not store personal health information (PHI), we maintain a formal incident response procedure to protect our users:
Our commitment: If we ever identify a security incident that may have exposed user data, we will notify affected users within 72 hours via email and a prominent site banner.
Incident Classification
Critical: Any potential exposure of user-submitted health data, API key compromise, or unauthorized access to processing infrastructure — triggers immediate investigation and 72-hour notification
High: Safety system failure (crisis detection bypass, disclaimer removal, CoT filter failure) — triggers immediate remediation and public disclosure within 7 days
Medium: Content drift, stale medical references, or degraded tool accuracy — triggers remediation within 30 days with updated audit documentation
Notification Procedure
Step 1 — Detection: Automated monitoring, user reports, and quarterly audits identify incidents
Step 2 — Assessment: Incident is classified by severity and scope within 24 hours
Step 3 — Containment: Affected tools are taken offline or restricted while investigation proceeds
Step 4 — Notification: For Critical incidents, users are notified within 72 hours via email and site banner. For High incidents, a public disclosure is posted to this page within 7 days
Step 5 — Remediation: Root cause is documented, fixes are deployed, and the incident is logged in our internal tracking system
Step 6 — Post-Incident Review: All Critical and High incidents trigger a full portfolio safety review to prevent recurrence
To report a security concern or potential breach, contact anchorwellpress@gmail.com with the subject line "Security Concern." We aim to acknowledge all reports within 24 hours.
Continuous Safety Monitoring
AI safety is not a one-time implementation — it requires ongoing vigilance and refinement:
Automated Safety Audits: Every 90 days, automated safety audits test for model drift, safety regression, and harmful outputs
Post-Launch Monitoring: Tools are monitored for content staleness, accuracy degradation, and emergent safety issues
Crisis Detection Testing: Regular adversarial testing ensures crisis detection mechanisms continue to function correctly
Regulatory Updates: When new state AI laws take effect, we conduct compliance reviews and update our systems accordingly
User Feedback: Safety concerns reported by users are investigated and acted upon promptly
Questions About Our AI Safety Practices?
We welcome questions, feedback, and reports of safety concerns related to our AI practices and ethical standards.
Subject Line: "AI Safety Inquiry" or "Safety Concern"
We aim to respond to all safety-related inquiries within 5 business days.
Our Commitment to You
AnchorWellPress is a health technology platform built on the principle that technology should serve humans, not the other way around. We are committed to developing AI tools that are safe, transparent, and respectful of your autonomy and privacy.
Every decision we make about our AI systems prioritizes your wellbeing and informed decision-making over metrics or profit.