EU AI Transparency Rules Arrive 2 August: Companion User Checklist
Final EU guidance turns AI disclosure into a practical check. What companion users can verify from 2 August—and what a label still cannot prove.
Auditing AI companions before they get personal.
Evidence-led reviews and updates on AI companions: clear pricing, data limits, and verified routes—without the hype.
Fresh reporting
View all stories →Latest stories
Consent is a product control: how companion apps communicate boundaries
Companion apps cannot treat consent as decorative dialogue. Boundaries should shape generation, memory, image handling, persona design, blocking and complaint systems, especially when a service enables adult roleplay or user-supplied…
Can you really leave an AI companion? Testing deletion, exports and retention language
A visible delete button is only the beginning of an exit test. A trustworthy companion service should distinguish account closure, conversation deletion, generated-media removal, model-training objections, export and backup retention.
Tested & Verified
Browse all reviews →
AI Companion Reviews
Dreamz.ai Review: Privacy, Gems, Billing and Safety
An evidence-led Dreamz.ai review of its documented AI chat, Gems, renewal questions, privacy boundaries and safer first-session checks.
Kupid AI Review: Privacy, Subscription and Account Controls
A source-backed Kupid AI review covering documented companion media, personal-data handling, subscription checks and account-exit questions.
Candy.ai Review: Privacy, Content, Billing and Controls
An independent Candy.ai review of documented chat and media scope, data use, content licensing, billing checks and safer trial boundaries.
AI CompanionsConnection,
Explore the beat
Connection,
simulated.
A clear-eyed look at the products designed to remember us, affirm us, and keep us coming back.
Explore the beatEU AI Transparency Rules Arrive 2 August: Companion User Checklist
Final EU guidance turns AI disclosure into a practical check. What companion users can verify from 2 August—and what a label still cannot prove.
Useful, not breathless
Start with the evidence
01 / PRIVACYThe AI companion privacy checklist
Five settings to inspect before your first conversation.
02 / METHODHow EmberGF tests products
Our repeatable protocol for claims, cost, safety, and control.
03 / DISCLOSUREHow this site makes money
Clear labels, independent verdicts, and no pay-to-rank lists.
04 / SAFETYBoundaries are a product feature
What consent-centered AI design looks like in practice.
Evidence desk
Memory claims need a repeatable test.
A published message limit is only a starting point. EmberGF checks recall, contradictions, unwanted invention and cancellation controls before turning a product claim into an editorial verdict.
Read the review method →
Meet Ember
A fox with a notebook—not a product to sell.
Ember is our recurring editorial correspondent: curious about how AI changes dating and intimacy, skeptical of easy claims, and always looking for the human consequence behind the interface.
How we report