AI disclosure

AI assistance is legible, limited and reviewable.

AI is a supporting layer for explanation, comparison and discovery — not the sole interface and not a substitute for sourced product evidence.

What is AI-assisted here

The labelled summaries and comparison verdicts on Digite360 are pre-generated: written once at ingest and stored, not produced while you read.

Imagery is disclosed on the image itself, not only in metadata. Where we hold no manufacturer photograph of a product, the card and the product page show a generated render carrying a permanent “AI-generated product render” label. Where a manufacturer photograph exists, its source is recorded instead. We would rather leave the slot empty than fill it with something we cannot attribute.

News illustrations follow the same rule: an abstract generated image is captioned as depicting its topic rather than the story, and a story’s own publisher preview is credited to that publisher.

The chat assistant

Digite AI is a conversational assistant available from the chat launcher on every page. It answers from the product catalog and news it is given for that question — retrieved by the server, not chosen by the assistant itself — and every reply is checked before you see it: it cannot reference a product or article outside the set retrieved for your question, it cannot link anywhere off Digite360, and sentences that assert a rating, price or comparison the underlying record does not support are removed before display. When a reply does not pass those checks, or the assistant cannot answer, you see a plain message saying so and a suggestion to use search instead.

There is no personalised recommendation engine. The assistant does not hold a profile of you, does not tailor answers to your history, and does not remember a conversation once you close it — each session starts from nothing.

Evidence boundaries

AI output may explain or summarise available records, but it must not manufacture ratings, prices, test results or source confidence. Unsupported claims are removed from comparison verdicts and chat replies alike before display.

Feedback and correction

Every chat reply carries a thumbs-up and thumbs-down control, and the correction process described on the corrections page covers AI-assisted content the same way it covers any other page.