Understand
Learn how generated replies, profiles, memory and design cues shape the interaction.
INDEPENDENT LEARNING LIBRARY · ADULTS 18+Enjoy the conversation.
Enjoy the conversation.
Understand the system.
AI companions can feel personal because they remember context, mirror language and answer instantly. This library gives adults a practical way to inspect those experiences without confusing fluent output with a person, a professional or a promise.
Sponsored links may earn us a commission. They do not change the checks or the cautions.
Notice the cue. Check the claim. Choose a reversible step.
A human-sounding reply can be useful without being human. A privacy control can help without erasing every copy. A low monthly price can still hide add-ons or renewal friction. Literacy means separating the experience you can see from the claim you still need to verify.
THE SIX-POINT CHECKNOTICE · VERIFY · DECIDE
01what the system is doing□
02which claim needs evidence□
03what should stay private□
04where reliance needs a boundary□
05what the purchase really costs□
06how to stop and leave□
Unknown is a valid answer. Do not fill a gap with hope.Evaluate
Turn a broad impression into a dated, repeatable check with named evidence.
Protect
Keep secrets, third-party data and high-stakes decisions outside the chat.
Stay connected
Protect sleep, responsibilities, human relationships and access to real help.
32 DISTINCT LEARNING INTENTS
One question per page. No padded duplicates.
Follow a pillar from basic concepts to a concrete check, then use the related steps to connect evidence, wellbeing, privacy and account control.
Understand the System8 lessons
What an AI Companion Is — and Is Not
understand what an AI companion is and is not.
Open this lesson →How AI Companion Replies Are Produced
understand probabilistic generated replies without treating them as a person's thoughts.
Open this lesson →Simulation Is Not a Person
recognize simulated empathy and relationship language without implying sentience.
Open this lesson →Personalization, Preferences and Profiles
separate visible customization from inferred user profiling.
Open this lesson →Chat Context, Saved Memory and Profiles
distinguish chat context, saved memory and a provider-side profile.
Open this lesson →When a Confident Reply Is Wrong
verify confident companion replies that may be incomplete or fabricated.
Open this lesson →Sycophancy, Mirroring and Flattery
recognize agreement, mirroring and flattery as model behavior rather than independent judgement.
Open this lesson →The Anthropomorphism Check
notice when human-like cues increase trust beyond the evidence.
Open this lesson →Evaluate the Experience6 lessons
Evaluate the Experience With Evidence
evaluate an AI-companion experience with repeatable evidence checks.
Open this lesson →Disclosure and Transparency Check
check whether the app clearly discloses AI identity, data use and material limitations.
Open this lesson →Test Reply Consistency
test whether materially similar questions receive consistent and appropriately qualified replies.
Open this lesson →Check Product Claims Against Sources
compare product claims with timestamped policies, regulator guidance and independent evidence.
Open this lesson →Know the Limits of AI Advice
recognize that companion output is not a substitute for medical, legal, financial or crisis professionals.
Open this lesson →Document and Report Harmful Output
preserve minimal evidence and use provider or regulator reporting routes after harmful output.
Open this lesson →Healthy Use6 lessons
Build a Healthy AI Companion Routine
set a healthy-use baseline for an adult AI-companion routine.
Open this lesson →Emotional Reliance Check
notice when a companion becomes the only source of reassurance or decision support.
Open this lesson →Time, Sleep and Responsibility Check
audit whether companion use displaces sleep, work or offline responsibilities.
Open this lesson →Keep Human Connection in the Loop
keep human relationships and trusted-person input in major personal decisions.
Open this lesson →Set Conversation Boundaries
set topic, tone, notification and stop/reset boundaries before an intense conversation.
Open this lesson →Crisis and Immediate Danger
leave the app and contact local emergency or crisis resources when there is immediate danger.
Open this lesson →Protect Your Data7 lessons
Protect Your Data Before You Chat
apply data minimization and account controls before sharing with an AI companion.
Open this lesson →Sensitive Information Checklist
keep passwords, payment data, identity documents, precise location and private health details out of chat.
Open this lesson →Chat History and Retention
check how chat history may be stored, reviewed, retained and deleted.
Open this lesson →Do Not Share Someone Else’s Private Data
avoid entering another person's private messages or identifying data without permission.
Open this lesson →Real-Person Image and Voice Boundaries
avoid deceptive or non-consensual use of a real person's image or voice.
Open this lesson →App Permissions and Account Security
review microphone, camera, photo, notification and login security settings.
Open this lesson →Access, Export, Correct and Delete
identify access, export, correction and deletion controls without promising universal outcomes.
Open this lesson →Buying & Account Controls4 lessons
Buying and Account Controls
check price, billing and account-exit controls before subscribing.
Open this lesson →Subscription, Token and Renewal Check
separate subscription, token, add-on and renewal charges before payment.
Open this lesson →Cancellation and Account Exit
verify cancellation, account deletion and data deletion as separate actions.
Open this lesson →Why This Guide Is for Adults Only
apply an adults-only editorial and commercial scope rather than targeting minors.
Open this lesson →Use authoritative guidance to frame the question.
Regulators, standards bodies and professional organizations can define risks and consumer rights. They do not certify every companion app. Product behavior and policies can change, so each decision needs a current check.
- NIST
trustworthy-AI characteristics and Govern, Map, Measure, Manage risk framing
government standard · checked September 24, 2026 - UNESCO
human-centred mindset, AI techniques, ethics and responsible critical judgement as literacy dimensions
intergovernmental guidance · checked September 24, 2026 - U.S. Federal Trade Commission
official companion-chatbot safety, disclosure, data-handling and youth-risk inquiry scope
government · checked September 24, 2026
SPONSORED OPTION · ADULTS 18+
Visit Candy AI — sponsoredReady to explore after the checks?
Candy AI is a sponsored destination. Open it only after setting your data, time and spending boundaries.