The facts about GenAI

• Accenture research shows that over half of people now question online content’s authenticity, eroding trust in brand interactions.¹ There is an implicit meaning that AI-generated (GenAI) content is not trustworthy.
• A study found that 91% of consumers believe brands should clearly label their use of AI.²
• Hootsuite found that 62% of consumers are less likely to engage with or trust AI-generated content, especially among older generations.³
• A 2024 HubSpot report says that 34% of marketers use GenAI for research, 41% use it to outline content and 46% use it to draft content.⁴
• 78% of US adults perceive AI-written news articles could lead to spread of misinformation,⁵ while newsroom executives fear that inaccuracy will increase.⁶ 

Why do we need brand standards for GenAI?

While AI can streamline efforts, support personalisation, create astounding visuals, analyse data, and automate posts, it’s essential to use AI to ‘enhance, not replace, your brand’s voice’.⁷

The Association for Intelligent Information Management (AIIM) believes that ‘trustworthiness is not a reasonable expectation to set for AI’. However, they do believe those using it have a responsibility to ensure it meets certain criteria.⁸

AI-generated Google search results suggested it was safe to put glue on our pizzas.

One of those criteria is ensuring the outputs are factual. CNET Money used automation to create news articles; but an article on compound interest included factually incorrect calculations that inflated the return value of a deposit.⁹ And AI-generated Google search results suggested it was safe to put glue on our pizzas.¹⁰

This is because AI tools are subject to hallucinations; creating information out of knowledge gaps and editing together points of information. It works from the data it was trained on – subject to human error and bias – which is why some language models identified secretaries and flight attendants as women’s roles, as well as gendering specific emotional states.¹¹

AI learning can also produce dangerous results, in that autonomous vehicles (self-driving) were found to be more likely to hit ethnic groups because there was not enough training data on pedestrians of varying skin colour.¹²

Audiences may decide to use an AI detector to verify content, but this also comes with problems. Some detectors were found to be ineffective at spotting the difference between human and AI content.¹³ When analysing content, AI searches for probabilities around perplexity (low vs high unpredictability) and burstiness (differences in sentence length and structure).

Human‑created content tends to have high perplexity and burstiness, meaning that rewrites of GenAI content may be essential to reduce repetition, add a more personal tone and ensure correct usage of words or phrases.

Don’t abandon GenAI altogether – but consider it’s uses

None of this means that GenAI has no use in content marketing – it’s helpful for generating ideas, providing a framework to build out, or even creating all new images and videos. It can make short work of statistical analysis, reporting, and tasks that benefit from automation, such as scheduling.

Audiences want authenticity – something they can trust to use in their lives, learn from, or be inspired by. Authentic content tends to be storytelling that doesn’t outwardly sell anything and employs a range of voices from all levels of a business (including customer voices).

For some brands, it’s community-driven content,¹⁴ where real and organic conversations based on interests and values thrive in online communities. Customers and individuals who feel involved are ‘more likely to advocate for the brand and share their experiences with others’.¹⁵

Ask yourself important questions, like whether the content provides original information, can be considered reliable, or adds value to a conversation.

When using GenAI, special care should be applied for brand safety. If two brands are using the same GenAI tool, they could end up producing similar content – which is bad news for rankings, as unoriginal content can impact search evaluation.

Google has provided handy guidelines on how to create helpful content; a quality that is prized in rankings. It means asking yourself important questions, like whether the content provides original information, can be considered reliable, or adds value to a conversation.

The E-A-T evaluation measures are now E-E-A-T – experience, expertise, authoritativeness and trustworthiness. Google recommends focusing on the who, how and why of content.¹⁶ Who created the content? Do they have a byline or author information that shows their background and expertise? And how was the content created? What was the process – an interview, secondary research? If it’s created through AI, this should be disclosed and why it was used.

Finally, you should explore why the content was created. If it’s to manipulate search rankings, it violates Google spam policies.

How should marketers respond in 2025?

If you want your content to be authentic and trustworthy, it’s a great idea to develop an AI policy for your business when using GenAI for content creation.

• Disclaimer or declaration – be transparent about how the content was created.
• Always fact-check for inaccuracy and bias.
• Employ quality controls – is the data verifiable, objective and complete? Is the writing of a high standard? Are you using credible resources, expert voices and research? Is the information unbiased and unlikely to lead to any form of harm if used by the audience?
• Maintain your brand values – if necessary, considering rewriting any generated content to instil your brand voice
• Keep it new, original and creativity – be wary of rehashing the same arguments; add your opinion into assets
• Is it helpful? Does the content answer questions your audience are asking or likely to ask? Does it show reasoning and provide a good explanation?

If you want support with content, or ways to rewrite and improve GenAI content assets, get in touch today. To receive the Five Financial Marketing Trends for 2025 report, contact tony.dickson@editionsfinancial.co.uk.