Zoho SalesIQ and the EU AI Act: does your chat widget tell people it's AI?
Zoho SalesIQ pairs live chat with two bot layers: Zobot (scriptable, can run on Answer-Bot NLP or OpenAI models) and the resource-driven Answer Bot. Both speak through the same float button and window as your human operators, across web, mobile SDK and messaging channels. Article 50(1) turns on a question SalesIQ's UI doesn't answer for the visitor: is the thing typing an operator, a script, or an LLM? If a bot answers first and the widget doesn't say so, the disclosure duty is unmet on every channel where that bot is live.
The rule itself is short. Article 50(1) of the EU AI Act requires AI systems that interact directly with people to be designed so users know they're dealing with AI, no later than the first interaction, in a clear and distinguishable way. The exception for cases where it's "obvious" is narrow: a natural-language customer-service bot doesn't qualify just because it has a robot icon. A line in your terms of service doesn't satisfy it either. And it's easy to enforce, because a regulator can open your site and start a chat.
Where the disclosure lives in Zoho SalesIQ
- The float button (#zsiq_float) and chat window, shared by operators and bots
- The bot's display name and image, set when you configure the bot for your brand (Settings > Brands > your brand > Bot tab)
- The first bot message, Zobot flows and Answer Bot both typically open the conversation
- Every deployment channel you enabled: website, mobile SDK, WhatsApp, Messenger, Instagram, Telegram, the duty follows the bot, not just the website widget
Your Zoho SalesIQ disclosure checklist
- Open your site in a private window as an EU visitor and start a chat: is it explicit you're talking to AI before or in the very first exchange?
- In SalesIQ, go to Settings > Brands > your brand > Bot tab: check which bot answers first and what name/avatar it presents, a human-style identity needs an explicit AI line in its first message
- Edit the bot's opening flow (Zobot builder or Answer Bot configuration) to state the AI plainly (e.g. "I'm the DisclosureProof answer bot, ask me anything, or type 'agent' for a human")
- If your Zobot uses the OpenAI/ChatGPT card, treat it as generative AI for disclosure purposes, the model behind the flow doesn't change the duty, but it weakens any 'it's just a menu script' argument
- Audit non-web channels: the same bot on WhatsApp or Instagram needs the same first-interaction clarity there
- Screenshot the first-interaction state per channel and save the bot configuration with a date
Watch out for
SalesIQ's flexibility is the trap: bots and humans share one visual identity, bots deploy to six-plus channels from one toggle set, and the Zobot can quietly graduate from menu-script to LLM by adding an AI card. Each step is a small config change; none of them re-prompts you to update what visitors are told. The result drifts toward an undisclosed AI conversation surface spread across channels nobody re-audits.
Common questions
Our Zobot is a simple button-flow script, is that even 'AI' under Article 50?
A pure decision-tree with canned replies is arguably not an AI system interacting with users, but the line blurs the moment NLP intent matching (Answer Bot engine) or an OpenAI card enters the flow, both are standard SalesIQ options. Given how cheap the fix is (one honest sentence in the opening message), disclosing is the safer position.
Do we need the disclosure on WhatsApp and Instagram too?
Yes, Art. 50(1) is about people interacting with an AI system, not about websites specifically. SalesIQ deploys the same bot to messaging channels, and the first-interaction notice has to reach visitors there just as on the web widget. Screenshot each channel's first exchange for your evidence file.