Navigating AI-generated content

70% of people can't spot AI deepfakes, leaving students vulnerable to hidden AI in advertising without proper AI literacy.

In this episode:

• A *Guardian* investigation revealed that 70% of people cannot detect *AI deepfakes*, making students vulnerable to hidden *AI in advertising*.
• Brands like *Once* and *Maket* are using undisclosed *AI generated content* and *AI-generated influencers* due to lower costs and fewer risks compared to human talent.
• Current *AI transparency rules* are lagging; the UK's *Advertising Standards Authority* doesn't explicitly prohibit undisclosed AI, although the EU will require labeling.
• Developing *AI influencers ethics* and critical AI literacy is paramount for educators to equip students to navigate a world saturated with AI-generated information.
• Educators must foster critical thinking skills, including basic *AI deepfake detection*, and discussions around *AI transparency rules* in the classroom.

Chapters:

• 00:00 — Cold open & welcome
• 00:30 — The Guardian's investigation into undisclosed AI in advertising
• 01:15 — Case studies: Once, Maket, and Ashle using AI-generated influencers
• 02:15 — Educational implications: Why AI literacy is crucial for students
• 03:00 — Why brands use AI: Cost savings and managing public image
• 03:45 — The blurring lines of authenticity and 'plausible deniability' in AI content
• 04:30 — Regulatory response: Which?, Lisa Barber, and the Advertising Standards Authority
• 05:15 — The urgent need for AI deepfake detection and AI transparency rules in education
• 06:00 — Cultivating human judgment and ethics over technological prowess

How can teachers equip students to identify *AI deepfakes* and *AI generated content*?
Teachers can equip students by fostering AI literacy as a critical thinking skill, teaching them to question AI outputs, biases, and intent, and discussing *AI transparency rules* in the classroom.

What are the ethical concerns surrounding *AI in advertising* and *AI influencers ethics*?
Ethical concerns include misleading consumers with undisclosed *AI generated content*, eroding trust, and the potential for manipulation when 70% of people cannot detect fake videos.

Are there *AI transparency rules* or regulations for brands using *AI-generated influencers*?
Currently, the UK's *Advertising Standards Authority* does not explicitly prohibit undisclosed *AI generated content*, though the EU's new Artificial Intelligence Act will require deepfakes to be labeled.

Featuring: Dan Fitzpatrick, Once, Maket, Ashle, Reality Defenders, Get Real Labs, Which?, Advertising Standards Authority, Lisa Barber.

Follow AI in Education with Dan Fitzpatrick for more on AI in education.

Transcript

SPEAKER_00 0:00

If this episode makes you think, please let us know in the comments and support us by subscribing and leaving a review. Thank you. Today we are exploring brands quietly using AI generated influencers to promote products on social media, an investigation uncovered by Sarah Marsh, the consumer affairs correspondent for The Guardian. What the investigation found was that a worrying 70% of people are unable to correctly identify all the real and fake videos they're shown online, meaning consumers are frequently misled by what they see. Now this isn't just about a bit of clever marketing, is it? This is about something deeper that touches on trust and transparency, and it has some really compelling implications for us in education. First up, let's talk about what this Guardian investigation actually uncovered. It seems that companies are increasingly turning to AI generated content that purports to show genuine customer experiences. The problem? There's often no clear indication that the people in these videos or images aren't real. In fact, some content creators are even signing non-disclosure agreements, making it harder to talk about this behind the scenes work. It creates a sort of plausible deniability, as one former celebrity manager put it. The report specifically highlighted a photo app called One Summent, for example. They had videos on Instagram showing a bride crying talking about how pleased she was to have used the app at her wedding. The whole thing felt incredibly authentic, very emotional, but analysis by cybersecurity companies, reality defenders and get real labs, which specialise in deep fake detection, suggested these were likely AI generated influences. When The Guardian reached out, once didn't respond. Then there was the Make It app which helps people design and plan housing projects. They had a video where a woman, again appearing to be AI generated, raved about an interior designer. Make it did respond to The Guardian, saying these AI generated influences were an experiment to test marketing concepts, not a core strategy. And a fashion brand called Ashel, based in Dubai, posted photos of a woman wearing their clothes that appeared to show an extra finger, a telltale sign of AI generation. They later removed those images, stating it was because the designs were no longer in their collection, not because of the AI aspect. This really gets me thinking about the kind of world our students are navigating. If 70% of people struggle to tell what's real and what's AI generated content, how are we equipping young people to be critical consumers of information? Particularly when it comes to AI in advertising. It underscores the absolute necessity of AI literacy, not as a technical skill, but as a collaborative reasoning ability. It's about thinking with AI, not just using tools and critically evaluating its outputs. We need to teach students how to ask what my m what does the AI know or not know here? What biases might be embedded? What's the real intent? Now you might be wondering why brands are going down this road. Clarissa Mansbridge, a former celebrity manager who creates these hyper-realistic AI influences for brands, shed some light on it. She said brands are attracted by the lower costs. Imagine, not having to pay $20,000 or $70,000 for a traditional photo shoot or deal with human influencers who might generate bad press or have strong personal opinions. She estimates that anywhere from 40 to 60% of content from some big brands is actually AI made, but those NDAs keep it quiet. It sounds like a classic case of outsourcing the doing, doesn't it? Brands are outsourcing the creation of seemingly authentic user-generated content, but they're not outsourcing the core strategic thinking of what resonates with an audience. Mansbridge argued that the authenticity of user-generated content was always about resonance, not about who made it. If the content reflects a real consumer truth about the product, it connects. AI, she suggests, just gives brands a smarter, more scalable way to get there. But this is where the line gets blurry, and it's a huge lesson for education. If we focus too much on just the product bus of learning and not the process or the performance box of understanding, we open ourselves up to this kind of plausible deniability. We need to design learning that cannot be faked, because it demands depth, care, and imagination. The real value is not in what the machine produces, but in how the student responds, how they critically engage, how they bring their unique context and judgment. This is exactly why we talk about assessment need in a cognitive stretch, asking if AI can complete a task without the student's unique perspective. If this discussion about AI and education is sparking your imagination, make sure you follow and subscribe to the podcast for more insights and conversations. So what are the regulators saying about all this? Which? The consumer group, along with their editor Lisa Barber, are calling for clearer rules. Barber noted their own investigation found 70% of people couldn't tell real from fake videos, which is a significant issue for consumer trust. But here in the UK, the Advertising Standards Authority, or ARSA, says there's nothing in their rules explicitly prohibiting brands from posting AI generated content without disclosing it. Their spokesperson explained that the use of AI itself isn't the issue they assess, it's whether the ad is misleading. The EU's new Artificial Intelligence Act coming in August will it require deep fakes and manipulated content to be labelled, but that legislation won't apply here in the UK. This puts the owners squarely on us, doesn't it? We can't wait for regulation to catch up. We have to empower students with the skills to navigate this world. This isn't just about telling them to be careful online. It's about cultivating a sophisticated understanding of how AI works, its limitations, and its potential for manipulation. It means teaching the principles of AI deep fake detection as a critical thinking skill, not just a technical one. We need to be having conversations in classrooms about AI transparency rules, not just for brands, but for our own use of AI in learning, because ultimately machines can compute, they cannot wonder, they cannot care, and it's those human qualities, judgment, wisdom, ethics that we need to protect and cultivate in our students more than ever. The core issue here isn't the technology, it's the AI influencers' ethics that underpin its use and the imperative for transparency in a world saturated with AI generated content. The future of learning demands we teach students not to outsmart machines, but to outthink them, equipping them with the critical faculties to discern, question, and ultimately create with integrity in an AI powered world. That's all for today. Thanks for listening.