What AI companions and sex chatbots mean
An AI companion is software designed to hold a conversation that can feel personal; a sex chatbot adds adult sexual or romantic content. Neither is a human partner. For adults aged 18 and over, the central questions are what information the service collects, whose privacy is involved, whether using it fits existing relationship agreements, and whether it remains helpful. A convincing conversational style does not establish confidentiality, accuracy or professional care. This guide explains boundaries and support without recommending services or providing explicit prompts.
A chatbot can produce affectionate language without being able to make a human agreement. The decisions that matter concern real people: the adult using it, partners whose information might be entered, and anyone represented in an uploaded photograph or generated image. The digital consent entry separates permission to create, keep and share material. Treat those as different choices rather than assuming that a private conversation makes everything inside it acceptable.
Read the privacy information before sharing
Look for the organisation operating the service, its privacy contact, the information it collects and the purposes for which it uses it. Ask whether conversations are retained, reviewed by people, used to improve models, or shared with other organisations. Check whether settings apply to earlier conversations as well as future ones. These are questions to investigate, not claims that every provider uses every practice. The ICO’s AI guidance explains why organisations need to consider data protection throughout their use of AI.
Do not confuse a character’s promise with the provider’s privacy notice. A generated reply saying a secret is safe is not reliable evidence about storage or access. Read the actual account controls and help pages. If the information is unclear, keep identifying and sensitive details out of the conversation. There is no need to disclose an address, workplace, medical history or intimate account of a relationship simply to explore whether a tool is useful.
Other people’s information needs particular care. Remove names and identifying details before asking a general question, and avoid uploading somebody else’s private messages or photographs. A partner agreeing that you may use an AI service has not necessarily agreed to having their personal story entered into it. Consider whether the same question could be asked without including their information at all.
Secure the account and the device
The NCSC recommends strong separate protection for email, software updates and two-step verification. Email security matters because an email account can provide access to other accounts through password recovery. Use available account security controls and review logged-in devices. An apparently private chatbot account can still be exposed through a shared screen, notification preview, forgotten browser session or someone with access to the device.
Check what appears in notifications and account emails, particularly on shared devices. Consider downloaded transcripts, screenshots and cloud backups separately from the original conversation. Keeping a copy may help you record a complaint, but it also creates another place where sensitive information can be found. The adult sexting safety guide explains related privacy choices. Security measures reduce some risks; they do not make a service confidential in the way a reader might expect from a professional appointment.
Age assurance and the Online Safety Act
The Online Safety Act’s provider-pornography age-assurance duty in section 81 applies across England, Wales, Scotland and Northern Ireland. It is in force: the commencement regulations brought that section into force on 17 January 2025, checked on legislation.gov.uk on 8 October 2026. Ofcom explains that services publishing pornographic images, video or audio, including through chatbots, must use highly effective age assurance to prevent children accessing that content. This is a provider duty, not proof that an adult service is safe or private.
Scope depends on the service’s functions. Ofcom identifies user-to-user services, search services and services publishing pornography; a chatbot meeting those definitions, or forming part of such a service, can fall within the rules. Do not assume that every standalone text conversation is regulated identically. Nor should an advertised age check be treated as official approval. The Online Safety Act guide explains the wider framework.
Before completing an age check, look for an explanation of who conducts it, what information is required and how it is handled. Submit information only through the genuine service process, never in response to an unsolicited message claiming to be support. An age check does not replace your own decision about whether a provider should receive sensitive personal information.
Human relationships and boundaries
Adults can have different views about romantic or sexual chatbot use. Discuss what matters rather than assuming a universal definition of fidelity. Questions may concern privacy, time, spending, sexual content or simulated conversations involving a recognisable person. An agreement should not require someone to surrender account access or disclose every private thought. The boundary entry helps distinguish a personal limit from control over another adult.
If a partner is uncomfortable, listen before debating whether the software is a real relationship. The concern may be secrecy, shared money or information about them rather than the label attached to the tool. The consent negotiation entry offers a way to discuss actual choices. A chatbot’s approving answer cannot settle a disagreement between people or authorise crossing a partner’s stated limit.
Emotional attachment and when to pause
Feeling attached to a responsive conversation is not, by itself, a diagnosis. Mind’s guidance acknowledges that people may feel comforted or understood by AI, while warning about inaccurate information, dependency and difficulties managing without it. Long-term effects and who is most vulnerable remain uncertain. Avoid treating either enthusiastic claims or alarming anecdotes as a prediction about every adult user.
Look at effects on everyday life: sleep, work, relationships, spending and the ability to stop. Consider a pause if conversations repeatedly leave you distressed, if you feel obliged to keep paying, or if the tool is replacing contact and support you want from people. Set a time or spending boundary outside the conversation itself, and arrange an ordinary activity or contact that does not rely on the service.
Do not use an adult companion as crisis support or as the final authority on health, law or consent. If a reply encourages harm, secrecy from trusted support, or disregard for another person’s refusal, stop engaging with that advice. Mind cautions that AI may give wrong or harmful information and may fail to recognise a crisis. Use qualified human help when distress persists.
Deepfakes, nudification and real people
A fictional character and an image made to resemble a real person raise different privacy questions. Do not upload someone’s likeness to create sexualised material without their agreement. Generated or altered imagery can still cause real harm, even if no original intimate photograph existed. The deepfake intimate images entry explains the term without describing how to create such content.
The intimate-images law hub covers offence definitions, UK nation differences, reporting and removal. The 2026 Act changes guide distinguishes enacted changes from commencement. Consult those dated legal explanations rather than assuming every provision applies everywhere at the same time. This page does not provide a separate offence summary or imply that a tool’s availability makes its use lawful.
If you are targeted, keep useful evidence such as the page address, account details, dates and threatening messages without circulating the abusive image. Report through the service’s abuse route and seek specialist help. Do not send further images or money in an attempt to obtain a promise of deletion. The Revenge Porn Helpline supports UK adults experiencing intimate-image abuse, including reporting and removal advice.
Leaving a service and asking about deletion
Before closing an account, find out whether cancelling payment, deleting chat history and deleting the account are separate actions. Ask the provider what happens to retained information and any copies you have downloaded. Keep confirmation of a cancellation or complaint somewhere secure. Do not assume removing an app deletes information held by the organisation.
The ICO provides current guidance on asking an organisation to erase personal data and on the circumstances and exceptions involved. Read that guidance before making a request: deletion is not an unconditional promise that every trace disappears immediately. If the provider’s answer is unclear, ask specifically about the information concerned and raise the issue with its privacy contact. The ICO is the appropriate UK source for information-rights questions.
Common questions
Is feeling attached to a chatbot automatically unhealthy? No. Consider its effect on everyday life rather than diagnosing yourself from the feeling alone. Persistent distress or difficulty stopping is a reason to seek support.
Does an age check guarantee confidentiality? No. Age assurance concerns access by children. Read privacy and security information separately before entering sensitive material.
Can a chatbot consent on somebody else’s behalf? No. A generated response cannot give permission to disclose a real person’s information, use their likeness or ignore their boundaries.
Is deleting the app enough? It may not close the account, cancel payment or delete retained information. Check those actions separately with the provider.
Can AI advice replace a GP or therapist? An adult companion is not a substitute for professional assessment. Verify health information through NHS sources and use human support for ongoing distress.
UK help
For concerns about personal information, use the ICO’s public guidance and complaint route. For intimate-image abuse, the Revenge Porn Helpline offers specialist support to UK adults aged 18 and over. Its role includes advice on reporting and removal; it does not replace a police investigation or long-term counselling. Report immediate danger to 999 and non-emergency crime through the relevant police service.
For distress, a GP can help assess what is happening and discuss suitable support. In England, NHS guidance directs people needing urgent mental-health help to 111 and its mental-health option or an urgent GP appointment. Elsewhere use the urgent mental-health route listed by your national NHS service. Samaritans offers listening support on 116 123 across the UK. If life is at risk or you cannot keep yourself or somebody else safe, call 999 or go to A&E. You do not need to prove that chatbot use caused a problem before asking for help.