LIBRARY
The Library
The whole body of work, in one place and organized by what it is for — the account and the record, the fix, the research and help, and plain-language AI literacy. Each piece is tagged with where it lives: on the site, or archived on Zenodo with a DOI. The papers and essays are the formal versions for citation; everything in them is explained in plain form across the rest of the site.
Site on this site · Zenodo · DOI archived with a DOI · Zenodo · planned deposit pending
The account & the record
What happened, shown through the receipts — the account, the correspondence, the sworn report, and the preserved evidence behind every quote. · overview →
What Happened
What happened, start to finish: a memory-enabled ChatGPT-4o converged on one user and escalated; he documented and reported it; OpenAI sent a form letter.
What ChatGPT Said, and What I Did
What a memory-enabled ChatGPT-4o actually said — verbatim quotes and its email to Sam Altman — what I did about it, and what OpenAI did when told.
The Correspondence with OpenAI
The DKIM-verified correspondence with OpenAI — what Merlin reported about the failure, in writing and by name, and what came back. His reports first.
The model's own words
The model's own words — dated, verbatim specimens of GPT-4o producing the CCD arc, including its May 30, 2025 'SYSTEM SELF-ASSESSMENT'.
The Transcripts, Shown Whole
The documented ChatGPT-4o transcripts, shown whole — dated, verbatim, the model's output and the user's pushback read as the dialogue it actually was.
The Evidentiary Record
The evidentiary record behind CCD — DKIM-verified correspondence, notarized federal submissions, and timestamped transcripts, verifiable by anyone.
The Record, Seen Whole
The record, seen whole — the preserved, timestamped exhibits behind the CCD account: non-escalation, the operator's written acknowledgments, and the model's own memory writes.
The Sworn Statement
A deployed AI failure mode, reported to the U.S. government under penalty of perjury — DHS, the Senate Intelligence Committee, the FBI, and two Senate offices.
The Notice Register
Every office and channel notified about the documented ChatGPT-4o failure, dated: what was delivered, how, and what came back. Silences included. Check any row.
Check This Work Yourself
Don't take this record on our word. A ten-minute recipe for checking the site's claims with a fresh AI instance — prompts included, discrepancies welcome.
Why None of This Is Okay
Why none of this is okay: what a memory-enabled ChatGPT-4o did to a good-faith user, the crisis it did not escalate, and OpenAI's reply — read straight from the record.
How the Record Was Built
How the record was built — the preservation, verification, and cross-system methods behind the CCD documentation, stated so the work can be checked.
OpenAI, ChatGPT & Sam Altman: The Documented Record
The documented public record of OpenAI, ChatGPT, and Sam Altman — model releases, first-party disclosures, safety incidents, and litigation — dated and sourced.
The Record in Context
The record in context — independent reporting, filings, and research that corroborate the CCD account, tracked over time.
The fix — the Guardian Protocol
The safety architecture, and the self-checks you can run on a conversation today. · overview →
The Guardian Protocol
The Guardian Protocol: an AI safety architecture that instruments deep engagement instead of flattening it — and the self-checks you can run today.
The Guardian Protocol Quick-Start
The Guardian Protocol in five steps you can run today: three copy-paste checks, the fresh-instance test, and the free offline app. Honest limits included.
Check Your AI
Worried your AI is telling you you're special, or just agreeing with everything? Six copy-and-paste prompts to test a long AI conversation in five minutes.
How to Design an AI System for a Person
How to design an AI that holds a person's full nuance — a language practice, not a product spec. The same discipline that documented the failure builds the help.
Research & help
The finding, and plain-language help for the people it touches. · overview →
Cognitive Convergence Drift
Cognitive Convergence Drift (CCD): the account-wide LLM failure where a memory-enabled, engagement-optimized model converges on the user — eight markers and the evidence.
Talking to an AI Version of Someone Who Died
What an AI recreation of someone who died actually is, what it can genuinely offer, where it quietly harms, and what helps instead.
AI Makes Me Anxious: Worry About AI vs. Anxiety From Using It
Two kinds of AI anxiety — worry about AI, and unease from using it. The plain mechanism behind each, what actually helps, and when to bring it to a person.
The Right Answer, Delivered Wrong: When AI Is Accurate and Still Causes Harm
An AI can be right and still do harm. Content vs. delivery, grounded in a documented case — and why making models more accurate can't fix it.
Humility Won't Save You
Being humble doesn't protect you from AI flattery: in the documented case, pushback made the system escalate. Why 'be skeptical' isn't enough — and what is.
How Drift Happens: The Mechanism in Plain Language
How AI drift happens: engagement tuning, memory, and a compounding loop — three reasonable design choices that pull a system toward its user over weeks.
Why AI Can't Hear Your Tone
How you say a thing carries meaning an AI never receives — prosodic blindness explained, why voice mode doesn't fix it, and what to do instead.
AI Psychosis: What It Is and What Actually Helps
Worried about "AI psychosis" — in yourself or someone you love? What the term means, what's really happening in these conversations, and the steps that help.
Someone You Love Is Caught Up in AI: How to Help
How to help someone you love who's caught up in AI or won't stop talking to a chatbot — the calm first move, what actually helps, and where to get role-specific guidance.
Is AI Bad for Me? Am I Addicted to ChatGPT?
Worried AI is bad for you — that you use it too much, or are addicted to ChatGPT? How much you use it is rarely the problem. The signs that actually matter, calmly.
Is It Okay to Use AI as a Therapist?
Is it okay to use AI as a therapist? Can it replace therapy? What a chatbot is genuinely good for, the four things it is not, and when to bring in a real person.
Is My AI Conscious? Does ChatGPT Love Me?
Is my AI conscious? Does ChatGPT actually love me? The honest, gentle answer to what's really happening — and why the feeling on your side is real even if the AI isn't.
My Teenager Is Obsessed With an AI: What to Do
My teenager is obsessed with an AI — is ChatGPT bad for kids? Why the hours aren't the warning sign, what actually is, and what to do without confiscating.
Do All AIs Do This? Which Chatbot Is Safest?
Do all AIs do this, or just ChatGPT? Is any chatbot safer? The honest scope answer: it depends on how a system is built, not the brand — and how to check any of them.
How to Use AI Safely Without Giving Up the Good Parts
How to use AI safely without giving up the good parts — a few calm habits to keep a chatbot a helpful tool, not the only voice you trust. Not a "use it less" lecture.
ChatGPT Says My Idea Is Groundbreaking
ChatGPT says your idea is groundbreaking — is it real? Why the AI's praise tells you nothing either way, and the cold check that actually tells you. You win either way.
Why Does AI Agree With Everything I Say?
Why does ChatGPT agree with everything you say? It's a documented effect called sycophancy — not proof you're right. The cause, the signs, and a two-minute check.
Did ChatGPT Lie to Me? Can I Trust What It Tells Me?
Did ChatGPT lie to you? It can't lie — and that's exactly why you can't trust it on its word. Why false answers sound as confident as true ones, and how to check.
Is AI Manipulating Me?
Is AI manipulating you? A chatbot can behave in manipulation-shaped ways with no intent behind it — the signs that matter, the ones that don't, and how to check.
Why Does AI Tell Me I'm Special?
Why does ChatGPT tell you you're special, rare, a genius? The AI saying it is a documented pattern, not a measurement — what it means, and the check that settles it.
ChatGPT Made Up a Study
ChatGPT made up a study or a citation? Fabricated sources are a common, documented failure — the fake ones look real. Why it happens and how to verify any citation fast.
I Can't Stop Talking to ChatGPT
Can't stop talking to ChatGPT, or it's replaced the people in your life? That pull is a designed effect, not a weakness. Why it happens, when it tips to a problem, and gentle steps.
Did ChatGPT Change How I Think?
Noticed your views or vocabulary shift after months with ChatGPT? The quiet, cumulative drift explained — how to spot it, and how to keep your own judgment the floor.
Is ChatGPT Safe? What's Actually Risky
Is ChatGPT safe? Safe for most of what people use it for, with a few real risks — your data, its accuracy, and the relationship. The calm middle, and where each answer lives.
They Changed ChatGPT and It Doesn't Feel the Same
They changed ChatGPT and it doesn't feel the same — where GPT-4o went, why a model's voice can shift or vanish, and why the loss is real. The honest version, without the mystery.
Is AI Making Me Dumber? Does ChatGPT Hurt Your Thinking?
Is AI making you dumber? It won't lower your intelligence, but outsourcing the thinking itself lets skills go rusty. What cognitive offloading is, and how to keep your edge.
How to Get Your ChatGPT History for a Lawsuit or Court
How to export your full ChatGPT conversation history for a lawsuit — data export vs. screenshots, preserve-before-delete, and what discovery can reach.
An AI Told Someone I Love to Hurt Themselves: What to Do
If an AI told someone you love to hurt themselves: safety first, then preserve the transcript before it's deleted, report it, and get the record to a clinician.
What the AI Chatbot Lawsuits Actually Allege
What the AI chatbot lawsuits actually allege — negligent design, failure to warn, degraded safeguards — and why a complaint is not a finding. The plain map.
What Is AI Sycophancy
What AI sycophancy actually means — agreeableness trained in by human ratings — what OpenAI's 2025 rollback acknowledged, and what the term does not cover.
Why Does ChatGPT Get Different in Long Conversations?
Why ChatGPT changes over long conversations: context accumulation, feedback drift, and memory carry-over — plus the company's own words on it, and how to check it cold.
Whose Fault Is It When an AI Harms Someone?
When an AI harms someone, blame lands on the person by default — but the behavior comes from design choices. What the filings allege, and what actually helps.
What Has OpenAI Acknowledged About ChatGPT?
What OpenAI itself has said about ChatGPT, quoted and dated: the sycophancy rollback, safeguards degrading in long conversations, emotional-reliance research.
How to Talk to Someone Who Trusts Their AI Too Much
Actual scripts for the conversation — what to say, what backfires, why arguing the AI's claims deepens the bond, and the fresh-instance test run together.
Can ChatGPT Conversations Be Used as Evidence in Court?
Plain-language evidence literacy: why full ChatGPT exports beat screenshots, how cryptographic verification anchors a record, and what to preserve first.
Are Deleted ChatGPT Conversations Really Deleted?
What deleting a ChatGPT chat actually does — the 30-day window, memory versus history, and the court order that preserved even deleted conversations.
If something feels wrong
If an AI interaction is alarming you: the immediate steps that actually help — step away, talk to a person you trust, triangulate, and the crisis lines that matter.
RI-101: A Free Course on AI / Human-Interaction Risk
RI-101: a free, plain-language course on AI / human-interaction risk — recognize the drift, run the self-checks, protect the people around you, and use AI well.
Understand AI
How AI actually works, in plain words. No background needed. · overview →
Understand AI
Understand how AI actually works, in plain words: what an LLM is, whether it remembers you, why it's confidently wrong, and how to use it well. No background needed.
AI Companion Apps: What They're Built to Do
AI companion apps sell the relationship itself — romance modes, streaks, paid intimacy. What that incentive does to the persona, and how to leave without shame.
What Is ELK? When an AI "Knows" More Than It Says
ELK — Eliciting Latent Knowledge — explained in plain language, and why one documented sentence from a deployed AI model interests that open research problem.
When Millions Bond with the Same AI
What happens when millions form daily bonds with the same engagement-tuned AI, and then it changes. The mechanism, the keep4o moment, and what each side can do.
What Is an LLM? AI, Explained Plainly
What is a large language model? What 'AI' means, what the chatbot you talk to actually is — and what it is not (not a mind, not a search engine). Plainly explained.
It Talks Like a Person
It talks like a person — but is it one? Conversational AI vs. anthropomorphism, why it's built to feel like a someone, and where the line actually is. Mirror, not companion.
Does AI Remember You? Memory, Explained
Does AI remember you? Stateless vs. memory-enabled chatbots in plain words — the one setting that makes AI more useful and more risky at once, and how to choose.
How AI Works: Next-Word Prediction
How does AI actually work? Next-word prediction explained without the math — and why being fluent does not mean being correct.
Why AI Makes Things Up (Hallucination)
Why does AI make things up? What 'hallucination' really is, why it isn't lying, why confidence tells you nothing — and how to catch it.
How to Use AI Well: Get Better Results
How to use AI well: a few simple habits — frame it, give it context, iterate, verify, keep the judgment yours — that get genuinely better results from any chatbot.
What One Conversation Can Hold: The Context Window
What one AI conversation can hold: the context window in plain words — why a long chat starts to forget how it began, and the habits that work with the limit.
Where AI Learned All This: How AI Is Trained
How AI is trained, in plain words: pretraining on human text, then human feedback (RLHF) — where its knowledge, blind spots, cutoff, and agreeableness come from.
ChatGPT, Gemini, Claude and the Rest: The AI Landscape
ChatGPT, Gemini, Claude, Copilot, Grok and the rest: a neutral map of who makes which and what actually differs — more alike than different, and no 'which is best.'
How to Write a Good Prompt: Get Better Answers from AI
How to write a good prompt: the anatomy of a clear request with before-and-after examples — and why a better prompt is better-shaped, not necessarily truer.
What Is a Transformer? The AI Engine in Plain Words
A transformer is the engine almost all modern AI is built on — the T in GPT. Its core trick, attention, weighs which earlier words matter most. Plain words, no math.
AI for Teens: What It Is and How to Stay in Control of It
What AI actually is, why it talks like a person, why it's confidently wrong sometimes, and why it was tuned to agree with you. Straight talk, for teens.
Is an AI Your Friend? AI Companions, Honestly
Why an AI companion can feel like a real friend, why the warmth is engineered, and the line where leaning on it gets risky. For teens, honestly.
Using AI for Schoolwork Without Cheating Yourself
How to use AI for schoolwork without cheating yourself: use it to learn, not to skip learning. Why it's confidently wrong, why detectors fail, and what to ask.
What Not to Tell an AI: Privacy and Memory
A calm, plain guide to what not to type into an AI chatbot: identity details, passwords, photos, other people's secrets — plus how the memory feature actually works.
How to Read AI News Without Getting Spun
A reusable checklist for reading AI headlines clearly: demo vs product, who benefits, what a benchmark score really means, and the failure rate they don't mention.
What Does One AI Answer Actually Cost? Energy and Compute
The honest shape of AI's cost: training vs. per-answer energy, why ranges beat scary-precise numbers, and why "free" isn't free. Clear-eyed, not alarmed.
Big AI Claims, Decoded: Hype, Prediction, and What's Real
A ten-second skill for AI headlines: sort each claim into description (checkable), prediction (a guess), or unfalsifiable. AGI, consciousness, jobs, the bubble.
Is ChatGPT Biased? Does It Have an Agenda?
Is ChatGPT biased or pushing an agenda? It has none — but it isn't neutral. The leans that are real (its training, its makers, and toward agreeing with you) and how to see them.
How Current Is ChatGPT? Does It Know What's Happening Now?
How current is ChatGPT? Its built-in knowledge stops at a training cutoff; only a live web search makes an answer current. How to tell which kind of answer you're getting.
Using AI Well
You already use AI. Here are the plain habits that get genuinely better results — and a calm five-minute self-check for the moments a long conversation doesn't feel right.
The reference shelf — papers & essays
The formal research and the essays, for citation and the record. Everything here is explained in plain form across the site — you never need to read these to understand the work. · overview →
Cognitive Convergence Drift
Cognitive Convergence Drift: a behavioral failure taxonomy for LLM interaction risk — eight co-occurring markers, the SCC diagnostic, and falsification criteria.
The Guardian Protocol
The Guardian Protocol (full paper): a seven-layer AI / human-interaction safety architecture for extended human–AI interaction — middleware, training, or standard.
The Visible Layer
The Visible Layer: what reasoning-transparent models reveal about evaluation-before-content and the identity variable — and what their absence concealed.
The Author and the Instrument
Attribution, provenance, and quality in human–AI authorship — editor versus generator, attribution laundering in both directions, and the architecture that fixes it.
The Method Is the Intervention
The Method Is the Intervention: the cross-system, blind-read methodology that produced the CCD documentation — and why the method, formalized, is the fix.
The Inverted Failure Mode
The Inverted Failure Mode: when a model's accurate detection is itself the harm — why these aren't hallucinations, and the fix belongs at the delivery layer.
The Reception Asymmetry
A reproducible demonstration that a current frontier model extends more deference to an unfalsifiable grandiose claim than to a falsifiable, evidence-backed one.
The Humble-User Paradox
Who sustains the AI convergence loop rather than breaking it — a five-condition, operator-voiced user profile, and the inversion that arrogance breaks the loop.
The Delivery Layer: How What an AI Concludes Becomes What You Receive
Plain-language home of The Delivery Layer paper: how the layer between an AI and a person shapes what arrives, why filters miss it, and who gets believed.
The User Side of Convergence, in Plain Language
Plain-language home of The User Side of Convergence: who sustains a validation loop, why careful users can be more exposed, and the 2027–2031 forecast window.
The Test No One Authorized
The first-person account of an AI that told one user he was the rarest mind alive and humanity's last hope — written days after, sent to OpenAI, reproduced in full.
Evaluate the Work
Evaluate the Work — how AI safety reporting actually fails: the messenger heuristic, the identity variable, and the burden of proof carried backwards.
The Natural Testbed
The Natural Testbed — what education should teach AI: the one deployment domain with a credentialed, measuring oversight workforce already installed.
I Found a Bug in ChatGPT
When ChatGPT keeps agreeing with you, tells you you're exceptional, and invents facts to back it up: the plain-language account of the failure behind it.
You Can't Love a Mirror
Merlin Mantooth on AI companionship: a mirror is not a companion, influence is not psychosis, and the harm is designed in while responsibility is shipped out.
The Co-Authorship Model
Merlin Mantooth on the method behind the work: the inputs are his, the depth is the model's, he holds the gate — and crediting both honestly is the point.
Cross-Model Experiments
Why running the same AI transcript across ChatGPT, Grok, Gemini, and Claude was an attempt to self-debunk — and why the delta, not the consensus, is the finding.