Can You Tell If You're Talking To An AI? Transparency, Deepfakes And The New Disclosure Rules Explained In Plain English For UK Businesses

Hard to tell - AI can now write, talk on the phone, and generate images and video convincingly enough that people cannot reliably distinguish it from reality  ·  Disclosure rules - New 2026 rules require chatbots to identify as AI, and realistic synthetic media to carry labels and watermarks  ·  Deepfake risk - AI-generated deception - fake voices, faces and content - is a genuine and growing threat to businesses and customers  ·  Plain English - This guide's commitment - no technical background required

It is getting genuinely hard to tell what is real. AI can now write like a person, talk like a person on the phone, and generate images and videos of people and events that never happened - convincingly enough that ordinary people, and businesses, increasingly cannot distinguish AI from reality. This is one of the most important and unsettling developments of the AI era: for the first time, we are losing the ability to assume that text, a voice, an image or a video is genuine, and that has profound implications for trust in every interaction.

We are writing this as an education-first guide because this is a topic every UK business owner needs to understand clearly, both for compliance and for protection, and because it moved from a theoretical concern to a live, practical one in 2026. New rules took effect requiring AI to identify itself: chatbots must now disclose that they are AI rather than passing as human, and realistic synthetic media must carry labels and watermarks so people can know when what they are seeing was AI-generated. These rules reflect a growing recognition that the inability to tell real from AI is a serious problem that transparency requirements try to address. For business owners, this raises practical questions that deserve clear answers, and this guide provides them in plain English.

The questions on every business owner's mind are practical ones: Do you have to tell customers when they are talking to your AI rather than a human? How do you protect your business and your customers from AI-generated deception - the deepfake scams, fake voices and fabricated content that criminals increasingly use? And what does the arrival of convincing, sometimes undetectable synthetic content mean for trust, both in your business and in general? This education-first guide explains, in plain English with no technical background required, the state of AI transparency and deepfakes, what the new disclosure rules mean, and what UK businesses actually need to do about both compliance and protection.

Do You Have To Tell Customers They're Talking To Your AI?

This is the compliance question business owners most often ask, and the honest plain-English answer is: increasingly yes, and in any case you should. The clear direction of the new rules is toward transparency - requirements that AI chatbots identify themselves as AI rather than deceiving people into thinking they are talking to a human. Whether a specific rule legally binds your particular business depends on the regulations that apply to you and how you use AI, which is worth getting proper guidance on, but the trend is unmistakable and spreading, and the safe, sensible and increasingly-required approach is to be transparent: let people know when they are interacting with your AI. Beyond compliance, there is a strong trust and reputational case for transparency - customers who discover they were deceived into thinking an AI was human tend to feel misled and lose trust, whereas being upfront that 'you are chatting with our AI assistant' is honest, increasingly expected, and rarely a problem when the AI is genuinely helpful.

The practical guidance, then, is to build transparency into how you use customer-facing AI as a default, rather than trying to pass your AI off as human. Make it clear when customers are interacting with an AI chatbot or AI phone agent; if you publish AI-generated images or video, label them appropriately; and treat disclosure as a matter of honest practice and trust, not just legal compliance. This is not onerous, and it aligns with where both the rules and customer expectations are heading. Businesses sometimes worry that disclosing AI will put customers off, but the evidence and experience suggest the opposite: customers value honesty and good service far more than they mind knowing an AI is involved, and the businesses that get into trouble are the ones caught deceiving, not the ones being upfront. Transparency is both the compliant path and the trustworthy one.

Protecting Your Business From AI Deception

The other side of the coin is protection, because the same technology that lets you use AI transparently lets criminals use it deceptively - and businesses are prime targets. AI-generated deception is a real and growing threat: deepfake voices that impersonate an executive to authorise a fraudulent payment, fake messages and content that appear genuine, synthetic media designed to deceive. The crucial and unsettling shift is that you can no longer assume a voice, a message, or even a video is genuine - the traditional cues we relied on to know we were dealing with a real, known person are now forgeable. This is exactly the vulnerability that AI-enabled fraud exploits, and it catches out businesses that still trust a familiar-sounding voice or a plausible message at face value.

The primary protection is not technical detection - which is unreliable against good fakes - but verification habits: confirming unusual, high-stakes or unexpected requests through a separate, trusted channel rather than trusting the voice, message or video that made them. If a call that sounds like your boss asks for an urgent payment, or a message that appears to be from a supplier asks to change bank details, the protection is to verify it through a known, independent channel before acting - because the voice or message itself can no longer be trusted as proof of identity. Building these verification habits into how your business handles consequential requests, and making your team aware that convincing AI-generated deception is now a real threat, is the most effective defence. In a world where authenticity cannot be assumed, verifying it for anything that matters is the essential new discipline - and businesses that adopt it protect themselves from a fast-growing category of fraud.

A Simple Checklist For UK Businesses

1. Disclose your customer-facing AI: make it clear when customers are interacting with your AI chatbot or phone agent, rather than passing it off as human.
2. Label AI-generated media: if you publish AI-generated images or video, label them appropriately, in line with the direction of the transparency rules.
3. Get guidance on your obligations: check which transparency and disclosure rules apply to your specific business and use of AI, and comply.
4. Build verification habits: confirm unusual or high-stakes requests (payments, bank-detail changes, sensitive information) through a separate trusted channel, never trusting a voice, message or video at face value.
5. Educate your team: make everyone aware that convincing AI-generated deception - deepfake voices, fake content - is a real, current threat, so they verify rather than trust.

Sources

1. European Commission - EU AI Act transparency provisions (chatbots must identify as AI; realistic synthetic media must carry labels/watermarks), enforceable from August 2026
2. Skycrumbs / AI Agent Store - August 2026 governance and transparency news
3. UK NCSC - guidance on deepfakes, synthetic media and AI-enabled fraud
4. ICO - guidance on transparency in AI systems that interact with people
5. BraivIQ - Batch 32 AI Cybersecurity Arms Race, Batch 34 AI Content Saturation and Batch 27 AI Governance articles (internal reference)