Mandatory AI Disclosure Rules and Your Website or App: A Guide for D2C Brands in Europe

Direct answer: If your D2C brand operates a chatbot, AI shopping assistant, or any interactive AI system on your European storefront or app, you are now required to clearly disclose to users that they are interacting with AI rather than a human. This is not a suggestion buried in a policy document — it is a legal disclosure obligation that has to show up in the actual interface, at the actual moment someone starts chatting, in language a shopper can understand instantly.

The trigger for this piece is a specific, dated development: as reported by Digital Strategy EC in August 2026, chatbots and interactive AI systems are now legally required to disclose that users are talking to AI. For D2C brands across Europe — many of which have spent the last two years bolting AI-powered chat widgets, product recommendation bots, and virtual stylists onto their storefronts with almost no thought given to disclosure — this closes a gap that most teams didn't know existed. We don't have a precise figure on what share of European D2C sites are currently non-compliant, and it would be irresponsible to invent one, but the general pattern is easy to reason through: AI chat adoption in ecommerce accelerated far faster than legal review cycles did, which means a large number of live storefronts are running assistants that never went through a compliance pass at all. This post walks through what the rule actually requires, why it lands especially hard on D2C brands specifically, and what changes in the practical, pixel-level design of your website or app.

What the AI Disclosure Requirement Actually Is

The core idea is simple even if the implementation isn't: any interactive AI system that a user might reasonably mistake for a human — a chatbot, a voice assistant, an AI-driven product finder, a virtual try-on advisor that "talks" — has to make it unambiguous, at the point of interaction, that the user is not speaking to a person. This is being enforced as a legal requirement under European digital and AI governance frameworks, not as a voluntary best practice or a UX nicety you can defer to a future roadmap item.

What makes this different from a typical "add a disclaimer" compliance task is where the obligation lives. It isn't satisfied by a line buried in your privacy policy or terms of service. It has to live in the interaction itself — the chat window, the greeting message, the voice prompt, the first thing a user sees or hears when the AI engages them. That's a UI and UX problem as much as it is a legal one, which is exactly why it belongs on the desk of whoever owns your interface, not just your legal counsel.

Why This Rule Exists Now

The reasoning behind mandatory AI disclosure isn't mysterious. Regulators have watched AI chat interfaces get conversational enough, fast enough, that ordinary users increasingly can't tell whether they're texting a support agent or a language model. When a shopper believes they're talking to a real person about a return policy, a sizing question, or a delivery delay, and it turns out to be an AI system whose answers may be generated rather than authoritative, that's a trust and consumer-protection issue — especially in commerce, where the conversation often precedes a purchase decision or a dispute resolution. Disclosure closes that gap by making the nature of the interaction visible upfront, not something the user has to infer or discover after the fact.

It also reflects a broader shift in how European regulators think about AI in consumer-facing products generally: the emphasis is moving from "is the AI accurate" toward "does the user understand what they're dealing with." Those are different questions. A chatbot can give a perfectly correct answer about shipping timelines and still fail the disclosure standard if the user never realized they weren't talking to a person. That distinction matters for D2C teams because it means accuracy audits and disclosure audits are two separate workstreams — improving your bot's answers doesn't touch your disclosure exposure at all.

Why This Matters Specifically for D2C Brands in Europe

D2C brands sit in an unusually exposed position relative to this rule, for a few concrete reasons.

First, D2C is a category built on direct trust between brand and customer — there's no retailer intermediary softening the relationship. That direct-to-consumer promise is precisely what an undisclosed AI chat undermines: if a shopper later realizes the "stylist" who recommended a product, or the "support agent" who handled their complaint, was actually an AI system that was never identified as such, the breach of trust lands on your brand name directly, with no third party to diffuse it.

Second, D2C brands have been among the most aggressive adopters of AI chat tooling precisely because it's cheap to deploy and scales customer conversations without headcount. Product recommendation bots, size-and-fit assistants, post-purchase support chat, and even AI-generated styling advice are now standard features on D2C storefronts, often added quickly through a third-party widget with default settings never reviewed for compliance. That speed of adoption is exactly what creates exposure now that disclosure is mandatory rather than optional.

Third, European D2C brands operate under some of the most active digital governance attention of any consumer market in the world right now. A rule reported through an official EU digital strategy channel isn't a niche technical requirement — it's the kind of obligation that shows up in audits, marketplace platform requirements, and eventually in how payment processors and ad platforms vet merchant compliance. Getting ahead of it is materially cheaper than retrofitting it after a complaint or an audit flags it.

Fourth, and easy to overlook, is the sheer number of places AI now shows up on a typical D2C stack. It's rarely just one chatbot anymore. A single storefront might run a customer support widget, a size-and-fit conversational tool, a post-purchase AI email or SMS assistant, and a native app with its own voice or chat layer — often each added by a different team, at a different time, through a different vendor. That fragmentation is exactly what makes disclosure compliance harder for D2C brands than for a business with one narrow AI touchpoint: there isn't a single place to fix, there's an inventory to build and then a design pattern to apply consistently across all of it.

What Changes in Practice for Your Website or App

This is where the requirement stops being a legal talking point and becomes a design brief. Concretely, here's what needs to change on a typical D2C storefront or app:

The chat entry point itself. The moment a user opens your chat widget, before they type anything, the interface needs to make clear this is an AI system. That's a first-message design problem: the greeting text, the avatar, the labeling around the chat icon (an "AI Assistant" label reads very differently from a name like "Sarah" with a human-looking headshot, which is exactly the kind of ambiguity the rule is meant to eliminate).

Persistent, not one-time, disclosure. A disclosure that flashes once and disappears doesn't satisfy the spirit of "clearly disclose" — the safer interpretation, and the one that holds up under scrutiny, is a persistent visual or textual marker that stays visible throughout the conversation, not just a first-message disclaimer a user can scroll past.

Voice and multimodal assistants need equivalent treatment. If your app uses a voice-driven AI assistant for order status or styling help, the same disclosure obligation applies — it just has to be delivered audibly and immediately, not buried in an FAQ page you assume no one reads.

Handoff moments need clarity too. Many D2C support flows escalate from AI to a human agent partway through a conversation. That transition needs to be disclosed just as clearly as the AI's initial identity — a user should never be uncertain, at any point in the thread, whether they're currently talking to AI or a person.

Anything that reads as human by design has to be re-evaluated. Human names, human avatars, typing indicators designed to mimic human response pacing — all of these were UX patterns built to make AI chat feel more natural. Under a disclosure requirement, several of these patterns need re-examination, because "feels human" and "clearly disclosed as AI" can pull in opposite directions if not designed carefully.

Onboarding and first-run experiences need a second look too. Many D2C apps now introduce new users to an AI stylist or shopping guide during account setup, before the user has any context for what they're interacting with. That's precisely the moment disclosure matters most, because a new user has no prior relationship with your brand's chat patterns to fall back on — the first impression has to be the accurate one.

None of this is a copy-paste fix. It requires rethinking the actual interaction design of your chat and assistant surfaces — which is precisely the kind of work covered under UI/UX Design & Branding: auditing every AI touchpoint across your storefront and app, redesigning the disclosure moments so they satisfy the legal requirement without wrecking conversion or brand voice, and making sure the visual language stays consistent with the rest of your brand rather than looking like a bolted-on legal notice.

How to Approach the Redesign Without Hurting Conversion

The instinct many teams have is to solve this with the blunt instrument of a giant "THIS IS A BOT" banner slapped across the top of the chat window. That satisfies the letter of the rule and damages the experience. A better approach treats disclosure as a design constraint to be solved elegantly, the same way accessibility or mobile responsiveness became constraints good design teams learned to absorb rather than resist.

Practically, that means auditing every AI-driven touchpoint on your site and app first — chat widgets, recommendation engines with a conversational UI, voice features, any post-purchase automation that talks back to the customer — and cataloguing which ones currently lack any disclosure at all versus which ones have something but it's weak or easy to miss. From there, the redesign work is about finding the disclosure pattern that fits your brand's visual identity: a subtly persistent label in the chat header, a distinct avatar style reserved only for AI assistants versus human agents, a first-message tone that discloses AI status while still feeling warm and on-brand rather than like a legal warning label.

Worth being deliberate about sequencing here too. Start with whichever touchpoint carries the most conversation volume and the most purchase-adjacent conversations — for most D2C brands that's the main storefront support widget, not a secondary feature like a styling quiz. Getting the highest-traffic surface right first limits your exposure fastest and gives you a tested pattern you can then extend to lower-traffic touchpoints without redesigning from scratch each time. It also gives your team a real answer if a customer, partner, or platform asks about compliance status before every surface has been updated — "our primary customer-facing assistant is compliant, and we're rolling the same pattern out across secondary features" is a defensible position; "we haven't looked at this yet" is not.

This is also a good moment to look at adjacent structural work that reinforces trust signals across your site more broadly. If you haven't reviewed your structured data recently, 13 Types of Schema Markup Every Site Should Use is worth revisiting alongside this audit — clear, well-labeled markup on your product and FAQ pages reinforces the same clarity-first principle that disclosure rules are pushing on your chat interfaces, and it's good practice to tighten both at once rather than treating them as separate projects.

It's also worth remembering that disclosure design isn't unique to ecommerce chat — it's a broader pattern in how any site presents automated or assisted experiences to visitors. The same principles that make a virtual assistant's identity clear on a D2C storefront apply, in a different form, to any business presenting an AI-assisted interface to the public, whether that's product guidance, booking flows, or support. Teams building out Website Development for Architecture and Interior Design Studios, for instance, are increasingly weaving AI-assisted project inquiry tools into client-facing sites, and the same clear-labeling discipline applies there too — the underlying UX principle travels across industries even when the product category doesn't.

What This Means for Your Metrics, Not Just Your Compliance Checklist

There's a legitimate concern that AI disclosure will suppress chat engagement — that once a customer knows they're talking to AI rather than a person, they'll disengage, ask fewer questions, or abandon the chat before it influences a purchase. That's a real design challenge, not a reason to under-disclose. The fix isn't to hide the disclosure; it's to make the disclosed AI experience good enough that customers stay engaged anyway — faster responses, better product knowledge, and a tone that doesn't feel like a canned script.

If your AI chat currently drives measurable assisted revenue, it's worth tracking how disclosure changes engagement and conversion once it's implemented, the same way you'd track any UI change against a baseline. If you're benchmarking paid or assisted performance around this rollout, What Is a Good ROAS? Benchmarks by Industry (2026) is a useful reference point for setting realistic expectations rather than assuming disclosure alone will tank performance — in most well-designed implementations, clear disclosure paired with genuinely useful AI responses holds engagement steady, because customers were never fooled for long anyway and often prefer knowing upfront.

Pricing Context: What This Kind of Work Typically Falls Under

An AI disclosure audit and redesign is scoped work, not a guess. Here's roughly how it maps to Scult's service tiers, depending on how many AI touchpoints your storefront and app currently run:

Scope	Typical Tier	What's Included
Single chat widget, straightforward disclosure redesign	Essential — $1,000	Audit of one AI touchpoint, disclosure copy and UI pattern redesign, implementation guidance
Multiple AI touchpoints (chat, recommendations, voice) across site and app	Growth — $2,000	Full audit across touchpoints, consistent disclosure design system, brand-aligned implementation
Complex multi-market D2C storefront with regional compliance variations	Enterprise — $4,000+	Full UX and compliance audit, custom disclosure design system, ongoing design support across markets

Key Takeaways

• Mandatory AI disclosure for chatbots and interactive AI systems is now a legal requirement in Europe, confirmed by Digital Strategy EC in August 2026, not a discretionary best practice.
• Disclosure has to live inside the actual chat or voice interaction — a policy-page mention doesn't satisfy the spirit or likely enforcement of the rule.
• D2C brands are especially exposed because of direct-to-consumer trust dynamics and the speed at which AI chat tools were adopted without compliance review.
• Handoffs between AI and human agents need their own clear disclosure moment, not just the initial greeting.
• Good disclosure design protects both compliance and brand voice — it doesn't have to look like a legal warning label.
• Audit every AI touchpoint on your website and app now, rather than waiting for a complaint or platform audit to surface the gap.

Getting AI disclosure right across your storefront and app takes a real UX pass, not a copy-paste banner. If you want help auditing your current AI touchpoints and redesigning them to meet the new requirement without hurting your conversion, book a meeting with our team.

Frequently Asked Questions

What counts as an "interactive AI system" under the new disclosure rule?

Any system that carries on a conversation with a user in a way that could be mistaken for a human — chatbots, AI shopping assistants, voice-based support bots, and conversational product finders all qualify. The common thread is two-way interaction that responds to a user's specific input rather than displaying static content.

Does this rule apply to a simple product recommendation widget that isn't conversational?

If the widget just displays suggested products without a back-and-forth chat interface, it's less likely to fall under the same disclosure obligation as a conversational assistant. However, if that same recommendation engine is wrapped in a chat-style UI where a user types questions and gets tailored replies, it likely does qualify, so the safest approach is to audit based on interaction style rather than assuming static-feeling features are exempt.

Do we need disclosure on every page, or just where the chatbot appears?

The disclosure needs to appear at the point of interaction — where the chat or voice assistant actually engages the user — rather than site-wide. A general privacy policy notice elsewhere on the site doesn't substitute for disclosure at the interaction itself.

What's the minimum wording that satisfies disclosure?

There's no single mandated script, but the safest practice is language that is immediate and unambiguous, such as clearly labeling the assistant as AI in its name, avatar, or first message, rather than relying on subtle or easily missed phrasing.

Can we still give our AI assistant a human-sounding name?

You can, but the name alone shouldn't be the only signal — pairing a friendly name with a clear "AI Assistant" label, distinct avatar styling, or an explicit first-message disclosure is the more defensible approach than a human name with no other indicator.

What happens if our chatbot doesn't disclose and a customer complains?

Consequences depend on enforcement mechanisms in each jurisdiction, but the general risk pattern is the same as other digital compliance gaps: complaints can trigger audits, and non-compliance discovered that way is more costly to fix under scrutiny than addressing it proactively now.

Is this only relevant to large D2C brands, or does it apply to small ones too?

The requirement is about the nature of the interaction, not the size of the brand running it. A small D2C store running a basic AI chat widget from a third-party vendor is just as subject to the disclosure obligation as a larger brand with a custom-built assistant.

Our chat widget is a third-party plugin — is the vendor responsible for compliance?

The disclosure obligation generally rests with the brand presenting the interaction to its customers, regardless of who built the underlying technology. It's worth checking whether your chat vendor offers disclosure settings, but the responsibility to configure and verify compliant behavior sits with you.

How is this different from GDPR consent requirements?

GDPR governs how personal data is collected and processed; AI disclosure is about being transparent that a user is interacting with an automated system at all, independent of whether personal data is involved. A chatbot can be fully GDPR-compliant on data handling and still fail an AI disclosure requirement if it doesn't identify itself as AI.

Does voice-based AI (like a phone or in-app voice assistant) need disclosure too?

Yes — the same principle applies regardless of medium. A voice assistant needs to identify itself as AI audibly, near the start of the interaction, rather than relying on a written disclosure the user can't see.

What if our AI assistant only handles simple FAQs, not sales conversations?

The disclosure obligation is tied to whether the interaction could be mistaken for a human conversation, not to how consequential the topic is. Even a low-stakes FAQ bot should disclose its AI nature if it's conversational in format.

How long will implementing this typically take for a D2C storefront?

For a single, straightforward chat widget, a disclosure redesign can often be scoped and implemented in a couple of weeks. Sites with multiple AI touchpoints across web and app — chat, voice, recommendation engines — typically need a longer audit-and-redesign cycle to keep the pattern consistent everywhere.

Will disclosure hurt our chatbot's conversion rate?

It can create a short-term dip if disclosure is handled bluntly, but a well-designed disclosure pattern paired with genuinely helpful AI responses tends to hold engagement steady, since most users adapt quickly once they understand what they're interacting with.

Should we redesign our chat avatar and greeting message ourselves, or bring in outside help?

Internal teams can absolutely lead this, but because disclosure interacts directly with conversion, brand tone, and legal risk simultaneously, many D2C brands bring in dedicated UX support to make sure the redesign satisfies the requirement without degrading the experience.

What's the difference between "disclosure" and "explainability" in AI regulation?

Disclosure is about telling users they're interacting with AI at all; explainability is a separate, broader concept about how an AI system's decisions or outputs can be understood or justified. This rule specifically concerns disclosure, not full explainability of how the AI generates its responses.

Do AI-generated product descriptions or marketing copy need the same disclosure?

The current rule as reported centers on interactive AI systems — conversational touchpoints — rather than static AI-generated content like descriptions or ad copy. That said, transparency expectations around AI-generated content are a broader, evolving area worth watching separately.

How do we audit our current site for AI touchpoints that need disclosure?

Start by mapping every place a user can have a back-and-forth exchange with an automated system — chat widgets, in-app assistants, voice features, conversational recommendation tools — and checking whether each currently identifies itself as AI clearly and persistently. This audit is the natural first phase of a UI/UX Design & Branding engagement.

What if our chat handles both AI and live human agents in the same window?

That's actually one of the trickier scenarios — the interface needs to clearly indicate which mode the user is currently in, and disclose the switch the moment a handoff happens in either direction, so there's never ambiguity about who or what the user is currently talking to.

Is there a standard icon or symbol for indicating "this is AI"?

There isn't yet a universally standardized icon the way there is for, say, accessibility symbols, which means brands currently have latitude to design a disclosure pattern that fits their visual identity, as long as it's unambiguous to the average user.

Does this rule apply to AI used only on the backend, like inventory forecasting?

No — the disclosure requirement is specifically about AI systems that interact directly with users in a conversational or assistant-like way. Backend AI that customers never directly converse with, like demand forecasting or fraud detection, isn't the target of this particular rule.

What if our AI assistant sometimes hands off to a human without the user asking?

That handoff should be disclosed just as clearly as the initial AI identification — the user shouldn't have to guess whether the assistant just changed from AI to human mid-conversation.

Are there penalties specifically for D2C ecommerce brands, or is this a general digital rule?

The rule as reported applies broadly to interactive AI systems, not to ecommerce specifically, but D2C brands are disproportionately affected because of how heavily the category has adopted AI chat tools for sales and support.

Can we satisfy this with a one-time pop-up when a user first visits our site?

A site-wide pop-up shown once on arrival is unlikely to satisfy disclosure for a chat interaction that happens later in the session — the safer interpretation ties disclosure to the moment of AI interaction itself, not a general site visit.

Does this affect how we should structure our chat widget's terms of service link?

Your terms of service can still reference AI use in detail, but it shouldn't be the only place disclosure lives — think of it as supporting documentation, not a substitute for in-context disclosure.

What's the risk of over-disclosing, like putting "AI" everywhere on the page?

Over-disclosure mainly risks diluting your brand experience and cluttering the interface rather than creating legal exposure. The goal is clear, consistent, well-placed disclosure — not maximal repetition of the word "AI" across every element.

How does this interact with the EU AI Act more broadly?

Disclosure requirements for AI systems interacting with humans are part of a broader European push toward AI transparency, of which this chatbot disclosure rule is one concrete, near-term application relevant to consumer-facing businesses like D2C brands.

Should our AI disclosure look the same on mobile app and desktop web?

The core disclosure principle should stay consistent across platforms, but the actual UI pattern often needs platform-specific adaptation — what works as a persistent header label on desktop chat may need a different treatment in a mobile app's more constrained screen space.

What if we're a UK-based D2C brand selling into the EU — does this still apply to us?

If you're selling to and interacting with customers based in the EU through AI-driven chat, the safest approach is to treat the disclosure requirement as applicable to those customer interactions, since the obligation is generally tied to who you're interacting with, not solely where your company is headquartered.

Will this rule expand to cover other AI features, like AI-curated product feeds?

It's reasonable to expect disclosure expectations to broaden over time as AI becomes more embedded in ecommerce experiences, though the current confirmed scope, per Digital Strategy EC, centers on interactive systems users converse with directly.

How do we train our support team on the new disclosure requirement?

Your support and CX team should understand exactly where AI handles conversations versus where humans do, and be briefed on how handoffs are now expected to be disclosed, so front-line staff aren't caught off guard by customer questions about it.

Does disclosure need to be translated for multi-market European D2C sites?

Yes — disclosure should be presented in the language the customer is browsing in, the same way any other customer-facing interface text would be localized, so a French-language storefront needs the disclosure in French, not just in English as a fallback.

What's a realistic first step if we haven't looked at this at all yet?

Start with a quick internal audit listing every AI-driven interactive touchpoint on your site and app, then flag which ones currently have zero disclosure versus weak disclosure, so you know the actual size of the gap before commissioning a full redesign.

Can our existing brand design system accommodate this without a full redesign?

In many cases yes — if your design system already has clear component patterns, adding a well-designed AI disclosure component is often an extension of the existing system rather than a ground-up redesign, especially for brands with a single primary chat touchpoint.

How does this affect our onboarding flow if we use an AI assistant to guide new users?

Onboarding flows that use a conversational AI guide need the same upfront disclosure as customer support chat — new users shouldn't be left assuming they're chatting with a real staff member during their first interaction with your brand.

What about AI-powered search bars that respond conversationally?

If your search feature responds in a conversational, assistant-like way rather than just returning a results list, it likely falls under the same disclosure logic as a chatbot, since the interaction pattern is what triggers the obligation, not the feature's label.

Should disclosure be visible to screen readers and assistive technology too?

Yes — disclosure needs to be accessible in the same way any other critical interface information should be, meaning it should be properly marked up so screen readers announce the AI identification, not just rely on visual-only cues.

Is a text-based disclosure enough, or do we need a visual indicator as well?

Text disclosure alone often satisfies the core requirement, but pairing it with a consistent visual indicator, like a distinct avatar or icon reserved for AI, reinforces clarity and reduces the chance a user misses a text-only label.

What if our chatbot occasionally makes mistakes — does disclosure reduce our liability for that?

Disclosure addresses transparency about who or what the user is talking to; it doesn't automatically resolve separate questions of liability for inaccurate AI responses, which is a distinct consideration worth reviewing with your legal counsel alongside the UX changes.

How often should we re-audit our AI touchpoints for compliance?

Given how quickly AI features get added to ecommerce stacks, a periodic audit — at minimum whenever a new AI-driven feature launches — is more reliable than a one-time compliance pass you never revisit.

Does this rule affect email or SMS marketing that uses AI personalization?

The disclosure rule as reported is centered on interactive, conversational AI systems rather than one-way personalized marketing content, so AI-personalized emails or texts aren't the direct target of this specific requirement.

What's the biggest mistake D2C brands make when implementing this?

The most common mistake is treating disclosure as a copy change alone — adding a line of text without rethinking the surrounding interaction design — which often results in disclosure that's technically present but easy for users to overlook or that clashes visually with the rest of the brand experience.

Can AI disclosure actually build customer trust rather than hurt it?

Handled well, yes — customers generally respond better to a brand that's upfront about using AI and delivers a genuinely helpful experience than to one that tries to disguise automation, since the disguise itself is what erodes trust once discovered.

Do we need legal sign-off before launching a redesigned disclosure pattern?

It's good practice to have legal review the final disclosure language and placement before launch, even though the actual design and interaction work is a UX-led process — the two should move in parallel rather than sequentially.

What if our brand uses AI-generated avatars that look photorealistic and human?

Photorealistic human-looking avatars for AI assistants are exactly the kind of pattern likely to draw scrutiny under a disclosure rule, since they actively work against the goal of making the AI's identity clear — this is a strong candidate for redesign.

How does this rule interact with app store guidelines for AI features?

Some app store platforms are independently tightening their own requirements around AI feature transparency, so it's worth checking your app store listing guidelines alongside the regulatory requirement, since the two can overlap but aren't identical.

Is there a difference in how this applies to B2B versus D2C interactions?

The disclosure principle applies to any interactive AI system regardless of business model, but D2C brands face more frequent, higher-volume customer-facing AI interactions than most B2B companies, which is part of why the practical impact is heavier for D2C.

What should our internal documentation include for this compliance area?

At minimum, a current inventory of every AI touchpoint, the disclosure pattern implemented for each, and the date it was last reviewed — this creates a clear audit trail if a platform, partner, or regulator ever asks how you're handling the requirement.

How do we balance a playful brand voice with a compliance-driven disclosure requirement?

Disclosure language doesn't have to sound clinical — a brand can disclose AI status in a tone that matches its personality, as long as the meaning is unambiguous, which is a design challenge best solved through careful copywriting paired with UI placement rather than treated as an either-or trade-off.

What's the first deliverable we should expect from a UX audit on this topic?

A well-scoped audit typically starts with a full inventory of AI touchpoints across your site and app, an assessment of current disclosure gaps, and a set of proposed disclosure patterns aligned to your brand before any implementation work begins.

Will this requirement keep evolving, or is it settled now?

Digital AI governance in Europe has been moving quickly, so it's reasonable to expect refinements and expanded guidance over time even though the core disclosure obligation for interactive AI systems is now in effect as of the August 2026 reporting.