Your Chat Is Private; Selected Character Images May Not Stay Private

“Private” can describe a conversation, an account identifier, a stored file or every copy of a generated picture. Those are different promises. EroLove AI’s privacy policy says images in your chat are visible only to you there. The same policy says staff may choose some generated character images for a public showcase, feed, external character listing or promotion. The published copy is not linked to your account, name, email or messages. That combination deserves a precise reading, especially when the scene is intimate.

Four different objects

The conversation: The policy treats messages and memory as sensitive account data. It says the relevant conversation content is sent to third-party AI model processors to produce replies, images or voice; the company says it does not send an email address or identity for that task. “Only you see the image in chat” therefore describes the user-facing chat view, not a claim that processing never involves a provider.

The generated image in your chat: It remains accessible there so the conversation can display it. The policy says the chat copy is private storage and is deleted with the chat or account, subject to its stated retention rules. We have not tested a deletion request, so this is the provider’s documented process.

A selected public copy: The team may choose a character image and write its own caption. The policy says a published copy belongs to the character, is detached from the requesting user, and does not include the user’s name, account, email, messages or other conversation details. A claim that EroLove publishes user identities with images would contradict the text we reviewed.

Copies after deletion: A selected public copy is not in the user’s data export and remains published after the chat or account is deleted. This is the point at which “delete my chat” and “remove every appearance of the character image” split apart.

Apply the distinction to a real choice

Suppose a fictional adult character produces a portrait after a deeply personal prompt. You may be comfortable with the company storing the image privately in your chat but uncomfortable with an unlinked copy in a public feed. The lack of an identity link reduces one exposure; it does not necessarily satisfy your preference that the image itself never leave a private setting. Conversely, a neutral character portrait might pose little concern to you. The policy leaves that judgment to the user before generation.

Ask these questions in order: Is there a real person or a private fact in the prompt? Would the character image still feel sensitive without the prompt attached? Do you need the ability to remove all copies later? Is this image worth creating under a permission that continues after deletion? If the last two answers are yes and no respectively, decline the image request. Text-only interaction or a different creative method may fit better.

A narrower, fair criticism

We criticise the gap between the intuitive phrase “private chat” and the separately disclosed public-image permission. We do not allege a data breach, unlawful publication, or proof that a selected image reveals its requester. The reviewed privacy text says the opposite about identity linking. We also cannot verify which images have been selected or how staff apply their caption process.

Read current policy language before relying on it; policies and product behaviour can change. If you believe a published image depicts a real identifiable person or is unlawful, the policy describes a report route. Preserve enough information to identify the item without forwarding the intimate image widely. A careful report is different from a public accusation unsupported by evidence.