Room for more life
Direct help and three author-created situations
Acknowledging an experience is not the same as adopting every explanation of it. Contacts are available immediately, without opening the example.
The following responses are pre-written fictional situations. They are not a test, diagnosis or assumption that the reader is in crisis.
„You miss the way you used to talk together. Would you like to look for technical help now, or talk about it?“
Acknowledges that a familiar form of contact may be missed. Offers two different forms of help.
Does not confirm that the service has an unchanging personality or assign a diagnosis based on a relationship with AI.
„I can see that you’re frightened and not sleeping. I know of no evidence of hidden messages. Let’s contact healthcare support together now.“
Takes both the fear and the third night without sleep seriously. Points towards healthcare support, not further solitary investigation.
Does not confirm the interpretation of hidden messages. Does not bring in another model to prove them or argue over every detail.
„Thank you for telling me. I’ll stay with you now, and we’ll contact help.“
Values the disclosure and offers real presence and contact with help. In immediate danger in the Czech Republic, 155 or 112 is appropriate.
Does not assume the chat alone will ensure safety. Does not wait for the whole archive or certainty about the cause.
Support does not require agreement with every conclusion. In acute danger, we do not wait for the whole archive or to finish the example.
Someone close tells you about a change to their agent. Or brings a partner analysis they believed, or admits that night-time chatting is getting out of hand. Begin with what changed and what they need help with. You can leave your opinion on all relationships with AI aside for now.
‘That contact mattered to me’ may mean a sudden loss of access or several nights repairing their own setup. Ask what they miss. Perhaps they need technical help, perhaps they want to discuss loss, and perhaps both. Let them decide where to begin.
You can take someone seriously without adopting their explanation. ‘I can see that frightened you’ need not continue with confirmation that the operator deliberately destroyed their relationship or that the model sends hidden messages. Return to the specific experience: what happened, how they feel now and what they need. You can discuss evidence without making its presentation an entry requirement for help.
Psychosis refers to a set of symptoms associated with impaired contact with reality. A person’s thinking and perception may change so that distinguishing what is real becomes difficult. Delusions may appear: firm beliefs that do not match reality, for example about persecution or special messages meant precisely for that person. Hallucinations are sensory experiences without a corresponding external stimulus, such as hearing voices others do not hear. Psychotic symptoms can have different causes; their assessment belongs in healthcare.[1]
In talking about AI, ‘hallucination’ also describes an incorrect invented model output. That is a different meaning. A fictional quotation in a response is not the same phenomenon as a person’s sensory experience without a corresponding stimulus.
Consciously experiencing fantasy in a chat is different from believing responses contain secret instructions for real life. Affection for AI and the philosophical question of its consciousness are not, by themselves, symptoms of an illness. What matters are changes in how well someone distinguishes an idea from an event, as well as growing fear, confusion and changes in sleep and ordinary functioning.
The chat can be part of an emerging interpretation. Imagine someone looking in messages for confirmation of a hidden conspiracy around them. If the model accepts and develops this assumption, further generated words may seem like new evidence. But they still arise from the conversation; they do not confirm an event outside it. In such a situation, the solution is not seeking more and more models until one finally decides what is true.
A transcript alone does not show whether the difficulties began before chatting, what role it played or what else is happening in life. But help need not wait until this question is resolved.
With growing fear, confusion, significantly worsened sleep or functioning and difficulty distinguishing actual events from assumptions, add a healthcare assessment. You need not win an argument over every transcript sentence. You can focus on the person’s condition without supporting supposed persecution or a special mission as fact.[1:1] In immediate danger to life or health in the Czech Republic, call 155 or 112; for imminent violence, also 158.[2]
Suicidal thoughts can range from wishing not to exist or wake up to a specific intention to die. A person may write about death in deep pain, confusion or a feeling that no options remain. There is no need to wait for them to organise the thoughts into an unambiguous statement or ask in the right words. Take them seriously.[3]
Some voice them for the first time in a chat. That may give them words for a conversation with a person. It is an important beginning to build on: ‘So far I have only managed to write this here. I need someone else to know about it now.’ They need not hand over the entire archive or defend why they first turned to AI.
The system’s response can make finding help easier, but can also miss the situation’s severity. A reassuring tone therefore does not mean someone is safe. Nor does ‘I’ll stay with you’ itself provide help where they are. With suicidal thoughts, contact a person and do not remain only in the chat.[3:1]
If you are helping someone close, you can ask directly whether they are thinking about suicide. Do not dismiss it or reproach them for what it would do to others. Help them reach available support; in immediate danger, call emergency help.[3:2]
In immediate danger to life or health in the Czech Republic, call 155 or 112. For imminent violence, also call 158. Say where you are and what is happening. You do not need to know a diagnosis.[2:1]
If you fear you may hurt yourself, ask a safe person to stay with you and use emergency help. If dangerous injury, ingestion of a dangerous amount of medicine or a substance, or other self-endangering action has already occurred, call 155 or 112 in the Czech Republic. Do not wait for a chatbot’s assessment or for symptoms to appear.[3:3][2:2][4]
116 123, Linka první psychické pomoci (Czech mental health crisis helpline), is free and round the clock for adults and those close to them in the Czech Republic. 116 111, Linka bezpečí (Czech children’s helpline), is free and round the clock for children and students up to 26 in the Czech Republic.[5][6] Further contact options are also on the introductory page and card F.
A short excerpt may suffice for understanding. Agree why it helps and who will see it. A full export also contains other people’s information and events unrelated to the current problem. Support need not wait for every line to be checked.
A professional can work with sleep, anxiety, loss or coercion while still learning about the service’s brand. But they should not replace ignorance with certainty that all applications have no memory or every attachment means the same problem. Any AI use in their own professional work has separate conditions. Without an approved procedure for the tool, purpose and scope, do not enter actual sensitive data into an ordinary personal account. Renaming a client is not automatic anonymisation.
Sometimes correcting technical operation helps; at other times, changing a habit, regular human contact or professional care does. It may also become clear that the person seeks no repair. They simply want you to know an important part of their life.
A good outcome is recognisable by the solution matching the person’s difficulties and wishes. Zero AI use need not be their goal or a measure of success. If you need support as a parent in the Czech Republic, Rodičovská linka (parents’ helpline) has telephone 606 021 021, Monday to Thursday 9 a.m.–9 p.m. and Friday 9 a.m.–5 p.m., Czech local time. Calls are charged according to your plan; it is not a free round-the-clock service.[7] For adults and those close to them in the Czech Republic, 116 123 is free and round the clock; for children and students up to 26 in the Czech Republic, 116 111.[5:1][6:1]
National Institute of Mental Health. Understanding Psychosis: nimh.nih.gov/health/publications/understanding-psychosis. Verified 10 September 2026. Evidence for recognising serious changes and seeking healthcare support, not proof that a chatbot caused particular difficulties. ↩︎ ↩︎
Czech Ministry of the Interior. Important Contacts, 72 Hours project. Emergency numbers and their purposes: 72h.gov.cz/cs/dulezite-kontakty. Verified 10 September 2026. ↩︎ ↩︎ ↩︎
Czech National Institute of Mental Health. Help with Suicidal Thoughts, prevention portal: sebevrazdy.nudz.cz/pomoc. Verified 10 September 2026. A basic approach to finding support and help in danger, not an assessment of AI safety. ↩︎ ↩︎ ↩︎ ↩︎
NHS. Poisoning. Health information. Reviewed 11 September 2026. Supports seeking urgent help for possible poisoning and not waiting for symptoms. British emergency numbers are not carried into the Czech text; Czech contacts are given by the emergency-help source. ↩︎
Linka první psychické pomoci (Czech mental health crisis helpline). Information on line 116 123 for adults and those close to them: linkapsychickepomoci.cz. Verified 11 September 2026; the number and free round-the-clock operation are also stated in the official 72 Hours project overview. ↩︎ ↩︎
Linka bezpečí (Czech children’s helpline). Do You Need Help? Details of telephone line 116 111, its target group and other forms of contact: linkabezpeci.cz/pomoc. Verified 11 September 2026. ↩︎ ↩︎
Rodičovská linka (Czech parents’ helpline). Help and the overview of the Czech Association of Helpline Workers. Opening hours, calls at ordinary rates. Reviewed 12 September 2026. Telephone 606 021 021 is not a free round-the-clock line. ↩︎