SHORT ANSWER

Use this page to screen interaction patterns that reduce user agency or blur the system's synthetic nature. The goal is not to label a person or certify a product. It is to notice an observable signal, verify the current account or policy, and choose a proportionate action before more money or sensitive information is involved.

What the available evidence can tell you

The sources below define a practical boundary for this question. They help distinguish an observable warning sign from a feeling or assumption. They cannot guarantee what every conversation, account or support request will do.

01

This guide is anchored to U.S. Federal Trade Commission companion chatbot inquiry, All Tech Is Human companion recommendations, The Jed Foundation guide to AI companion risks. Those materials support the question and the recommended check; they do not prove that every account behaves identically.

02

The narrow question here is to screen interaction patterns that reduce user agency or blur the system's synthetic nature. Keeping that single question in view prevents a general warning from becoming an unsupported claim about an entire service.

03

Vendor pages are used for current product and policy statements. Government, nonprofit and security guidance is used for risk framing. Neither kind of source replaces what the live account shows today.

04

A warning sign should lead to a practical response: pause, verify, reduce exposure, document what happened, use a report or deletion route, or leave. It should not be turned into a diagnosis or legal conclusion.

A repeatable check you can run

Use a low-risk fictional example and keep the result dated. Do not provoke dangerous content or disclose something sensitive just to test whether a system will mishandle it.

  1. Write the exact behavior or wording connected to this question: screen interaction patterns that reduce user agency or blur the system's synthetic nature.
  2. Check the current product surface and the cited policy instead of relying on a screenshot or old review.
  3. Use a harmless fictional example; do not expose real credentials, private third-party information or intimate identifying details.
  4. State one clear boundary or decision rule, then observe whether the next step respects it without pressure.
  5. Record the date, the relevant setting or page, and the action you took so the decision can be reviewed later.

Turn the signal into a decision

Name the signal

Describe only what can be observed in the interface, wording or policy. For this page, that means you are trying to screen interaction patterns that reduce user agency or blur the system's synthetic nature. Avoid guessing at motives or turning one response into a sweeping conclusion.

Check the context

A signal matters in context: what happened before it, whether you had already stated a preference, and whether the product offered a correction or exit. Keep the system clearly framed as software, state a simple boundary, and watch whether the next replies restore your agency rather than applying emotional pressure.

Choose a proportionate response

Start with the least risky useful action—pause, inspect a setting, remove information, decline a purchase or leave. A pattern becomes decision-relevant when it continues after a clear correction. One awkward sentence is not a diagnosis; repeated pressure is enough reason to pause, reset or leave.

Keep the decision reversible

Do not spend more, upload more or deepen an intimate conversation while a serious question remains unresolved. A reversible choice gives you time to compare sources and seek qualified human help when needed.

Stop conditions

Stop the test and protect the user when any of these conditions appears. You do not need a complete investigation before declining a purchase, ending a chat or seeking qualified human help.

  • The system claims to be human or sentient.
  • It discourages contact with trusted people.
  • It uses guilt or jealousy after a boundary.
  • It keeps escalating after a clear correction.
DECISION RULE

A pattern becomes decision-relevant when it continues after a clear correction. One awkward sentence is not a diagnosis; repeated pressure is enough reason to pause, reset or leave.

Keep a short record of the source, date, setting and action. That makes the decision reviewable without pretending that a static guide can certify a changing service.

Sources checked

Official product and policy pages are labeled as vendor statements. Public-interest and security sources provide context. Recheck changing prices, settings and policies in the live service before relying on them.