Sector Briefing 04 · Human Dignity · 5-minute briefing
The Right to a Private Mind.
Privacy is not about hiding what you've done. It's about protecting who you are becoming. If you cannot think in private, you cannot be free.
This briefing makes one argument: the four risks below do not depend on the technology's weaknesses. Each grows with capability, because each runs through observation, inference, and influence, the things capable systems do best. Our tools have become part of how we think. Reading them is reading us.
Are you the user, or are you the training data?
The Risks
Risk 01 Holds at any capability
The Architecture of Total Capture
You have a right to your own internal monologue. You have a right to wonder about something without an algorithm using that data to sell to you or influence you. The architecture of total capture watches the drafting phase of thought itself: what you almost said, what you deleted, what you hesitated over.
Why a better system makes it worse: capture completeness scales with capability. The private workspace shrinks as the observer improves.
Dive Deeper
The Architecture of Total Capture · The Era of Cognitive Capture · The Extended Mind (Clark & Chalmers, 1998)
Risk 02 Holds at any capability
Ambient Surveillance
AI wearables put an observer in the room for every conversation, including the ones no one else agreed to have recorded. What they collect is reaction, attention, and mood: the raw material of a psychological profile, gathered from people who never touched the device.
Why a better system makes it worse: better sensors and better inference make the ambient layer complete. The wrong arrives before the law does.
Dive Deeper
Risk 03 Holds at any capability
Synthetic Media as Coercion
Fabricated media does not need to deceive anyone to do harm. Synthetic sexual imagery coerces, humiliates, and controls, and the target carries the burden of a fabrication she never consented to exist. This is a cognitive security problem before it is a content problem.
Why a better system makes it worse: fidelity is the weapon. The harm does not require anyone to be fooled.
Dive Deeper
Synthetic Media Is a Cognitive Security Problem · The Liar's Dividend
Risk 04 Holds at any capability
Inference Without Consent
A pattern of questions lets a system infer a medical condition, a mental state, or a belief the user never disclosed. The inference is the product. It is built from ordinary use, and there is no consent step anywhere in the pipeline.
Why a better system makes it worse: the profile gets more accurate, and consent never enters the pipeline.
Dive Deeper
AI Algorithms Can Read Your Mind · ChatGPT Ads and User Surveillance
1
Of the consumer-device companies whose privacy policies were reviewed, the number that explicitly encrypt biometric data beyond even their own employees’ reach. Magee, Ienca & Farahany (2024), Neuron.
48%
Share of users in a 55-account audit who did not know their AI’s memory feature existed. Reported in AI Memory and the Barnum Effect.
The Capture Pipeline
From private thought to external influence, in four steps.
01 You Think
Drafts, queries, hesitations, deletions
02 It Watches
Pre-upload scanning, sensors, session logs
Observed03 It Models You
Inference: confidence, mood, susceptibility
Inferred04 It Steers You
Targeting, ranking, nudging, denial
The pipeline runs on ordinary use. No breach is required, and no consent is asked.
Required Protocols
01 / Law
Cognitive Rights
Mental privacy established as a human right, distinct from data privacy.
02 / Right
Opt-Out of Inference
The right to use software without it building a psychological profile of you.
03 / Duty
Algorithmic Transparency
A system that is nudging your mood or opinion discloses that intent immediately.
Where This Sector Sits in the CPIA
Every domain touches the person. Domains 01, 02 and 03 carry the rights weight, and the questions below are the ones a user is entitled to ask of any system that watches her think.
Can you use the system without it building a psychological profile of you?
Can you see, contest, and delete the inferences made about you?
Does the system disclose when it is optimizing for your engagement rather than your interest?
What is retained from the drafting phase: the things you typed and deleted?
Does anything you did not choose to share leave the session?
Who else can reach the model of you the system holds?
The Definitions · Cognitive Privacy & Security
The two terms this project is built on are defined, dated, and maintained here. Start with them; the research follows.
What Is Cognitive Privacy · What Is Cognitive Security · The Era of Cognitive Capture · All Research
Cite this briefing
Cook, T. (2026). The right to a private mind: AI risk to human dignity. The Cognitive Privacy Project. cognitiveprivacyproject.org/human-dignity
For advisory inquiries or research collaboration: timothy@cognitiveprivacyproject.org