Olark and the EU AI Act: does your chat widget tell people it's AI?
Olark's core identity, for over a decade, has been human live chat — the brand promise was "real people, not bots." Aiden, its newer AI Assistant available on the Pro plan, turns your knowledge base into an automated agent that answers before any operator sees the conversation. That history is the Article 50(1) risk: sites running Olark since long before Aiden existed often still carry welcome copy and widget branding built for a human-only chat, which doesn't update itself the day someone switches Aiden on.
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 simply open your site and start a chat.
Where the disclosure lives in Olark
In a typical Olark setup, the surfaces that can carry (or fail to carry) the AI disclosure are:
- The chat box launcher and default greeting message, often unchanged from the pre-Aiden, human-chat era on long-running Olark installs
- Aiden's presented identity when it answers — name, avatar, and whether it's visibly distinguished from a live operator
- Aiden's first automated reply, generated from your Knowledge Base before any operator sees the conversation
- The handoff moment when a visitor's request to "speak to an agent" routes from Aiden to a live operator
Your Olark disclosure checklist
- Open your site in a private window and start a chat: before Aiden's first reply, is it clear you're talking to AI rather than a human operator?
- Check whether Aiden is presented under its own name/avatar or blended into general "chat with our team" branding — Olark's classic pitch is human chat, so this needs a deliberate check, not an assumption
- Add an explicit AI statement to Aiden's opening reply, since older Olark greeting text was written for a human-only widget and won't say it for you
- Confirm the "speak to an agent" handoff clearly tells the visitor a person has joined, not just that replies changed in tone or speed
- Test both the pre-Aiden welcome message and Aiden's in-conversation replies — fixing one doesn't automatically fix the other
- Screenshot the first-interaction state and export your Aiden/Knowledge Base configuration with a date
Watch out for
Olark's decade-plus brand promise was live chat with real operators, and Aiden was layered on later as a Pro-plan add-on. That history is exactly the risk: teams that installed Olark years ago, running copy that still says "chat with our team," don't get automatically updated the day Aiden starts answering first. The fix is a deliberate audit, not a default you can trust.
Common questions
We only use Aiden after hours or during overflow — does that change anything?
No — disclose wherever Aiden can be the first responder, which for many stores includes exactly those windows. If Aiden can open a conversation, its first message needs to say what it is.
Olark markets itself as "real people" — do we need to change that copy?
Not necessarily change it, but make sure it's still true for the conversations it describes. If Aiden now answers first on some or all chats, blanket "real people" copy next to an AI-led widget is the kind of inconsistency a complaint would zero in on.
Check it in one scan.
DisclosureProof opens your site the way a regulator would, triggers your Olark widget, and records whether the AI disclosure actually appears — with timestamped evidence either way.
Scan your site free