Yes — but regulation is catching up fast. Here is the honest 2026 legal landscape and what to watch.
Most jurisdictions in 2026: legal for adults, with restrictions on minor access, marketing claims, and data handling. The EU is the most regulated under the AI Act and GDPR. The US is state-by-state, with several states requiring age verification on companion AI with romantic features. China has its own rules requiring algorithm registration. Japan and South Korea are more permissive. Russia restricts foreign AI services more broadly.
undefined
Clear age verification on tiers with romantic content. Transparent data handling (preferably sovereign). No marketing claims of sentience or feeling. Crisis-routing for mental-health emergencies. Clear in-app communication that the AI is an AI, not a person. Removable and exportable user data. Compliance with the AI Act for EU users.
Targeting minors with companion AI that has romantic features. Claiming an AI loves you or feels for you in advertising. Mishandling intimate user data via third-party LLM pipelines without consent. AI companions used to manipulate vulnerable users into purchases or harmful behaviour. The norms are hardening fast.
Luna is positioned as a warm AI companion, not a romantic AI girlfriend. We do not claim Luna loves you, does not have romantic / NSFW modes, and does not target users under appropriate age tiers.
Sovereign infrastructure means the data-handling concerns that triggered Replika's GDPR fine do not apply by construction.
Honest about what she is. No marketing of sentience. Compliant with EU AI Act guidance.
Unlikely across the board; very likely in specific cases of overclaim, harm, or targeting minors. The category is legal; specific bad actors are being removed. The mature category in 2027-2028 will look more like dating apps today — regulated, with clear rules, but legal.
No, generally. The legal action in this space targets operators, not users. Use of legal AI products is a personal choice not subject to legal risk in most jurisdictions.
Generally not on work devices — your employer's AUP may prohibit it, and intimate content at work is rarely a good idea. On personal devices and personal time, it is your call.
Illegal in many jurisdictions and rapidly being criminalised in others. Creating non-consensual deepfake AI partners of real people is one of the clearer harms in this category and is being addressed in law. Avoid any product that offers this.