AI Generated Graphics for Social Media: What the Research Actually Says
There is a conversation happening at the moment about AI generated graphics and posters on social media.
Some people think they save time. Some people think they all look the same. Most of it is opinion, and opinion is fine, but it does not help you decide what to do on a Tuesday night when you have a post to get out and no design skills.
So rather than add another opinion, here is what the research says. The good bits and the awkward bits.
First, some context on who is actually using it
AI use in UK business has moved fast.
The British Chambers of Commerce found in March 2026 that 54% of UK SMEs are actively using AI, compared with 35% in their 2025 survey, 25% in 2024 and 23% in 2023.
Marketing is the number one use case. The government's own DSIT research found it was named by around 72% of small business adopters.
On the professional side, Canva's 2026 marketing report found 97% of marketing leaders use AI in their daily creative work, and 99% plan to spend more on it this year.
So this is not a thing that a few people are doing. If you have thought about using AI to make a graphic, you are in the majority.
The case for using AI graphics
Speed and cost. This is the clearest benefit and the one the data supports most strongly. In DSIT's research, around 75% of AI adopters reported productivity gains. Lloyds put the figure even higher at 87%. For a business owner doing her own design at 10pm, that time is not a nice to have.
Access. If you cannot use design software and cannot afford to pay someone who can, AI tools have lowered the barrier considerably. The FSB found that 46% of small firms say a lack of knowledge is what holds them back with AI, which cuts both ways, but for design specifically, the skills gap that used to stop people has shrunk.
Volume. Canva's research found 68% of marketing leaders say AI has increased the number of marketing-influenced decisions their teams can make. More output, more testing, more chances to find what works.
Consumers are not universally against it. Canva found 68% of consumers say they do not mind AI in advertising if it makes the ad more helpful or relevant to them. The Harris Poll UK made a similar point in April 2026: consumers are not rejecting AI outright.
That last point matters, because the online conversation often suggests audiences are furious about AI. The research shows something more measured than that.
The case against, or at least the caveats
Trust drops when people can tell. This is the most consistent finding across the studies I looked at.
The Harris Poll UK found that seven in ten consumers trust adverts made entirely by humans. That falls to four in ten when AI tools are involved, and to two in ten when the advert is generated by AI alone.
Klaviyo and Datalily surveyed 8,000 consumers across eight countries including the UK in December 2025. Only 7% said visible AI generated marketing made them trust a brand more. 31% said it made them trust the brand less.
A 2026 Gartner survey found 50% of US consumers would prefer to give their business to brands that do not use generative AI in customer-facing content.
"Too perfect" is now a warning sign. The Harris Poll UK found that the things which make people suspect an image is AI are images that look too perfect (51%), images that feel unrealistic (43%) and small inconsistencies (42%).
That is a reversal of how design used to work. Polish used to signal effort and money. Now it can signal automation.
Detection is uneven and age-dependent. 46% of consumers say they are confident they can spot AI generated imagery. Among 18 to 24 year olds that rises to 72%. Among the over 65s it drops to 21%. So how much this matters depends heavily on who your customers are.
Sameness is a recognised problem, not just a moan. Canva's report found mentions of "AI slop" in media monitoring data had increased ninefold, and 41% of marketing leaders now name it as a real challenge in their own work.
The results are not showing up in revenue yet. DSIT found that while adoption and productivity gains are widely reported, only around 12% of AI-using businesses report increased revenue so far.
The legal bit that rarely gets mentioned
This is the part I see almost nobody discussing in the graphics debate, and it is worth knowing if you are using AI images to sell things.
Ownership is not straightforward in the UK. Under UK copyright law, computer-generated works sit in an unusual position, and there is live legal debate about whether a person who typed a prompt counts as the author. Law firms writing on this are consistent in saying the position is unsettled.
Your tool's terms matter more than you think. Different platforms grant different rights over what you generate. Some free tiers do not include commercial use at all. Legal commentators repeatedly advise checking the terms before using generated images in anything commercial.
Accidental similarity is a real risk. Because these tools learn from existing images, output can end up resembling protected work. Several UK law firms point out that liability usually sits with the person who published the image, not the tool.
Advertising rules still apply. The ASA and CMA regulate advertising in the UK regardless of how the creative was made. If an image creates a misleading impression of your product, the fact that AI produced it is not a defence.
None of that means do not use it. It means read the terms of the tool you are using, and do not use generated images to show a product or a result that does not exist.
What the platforms are doing
Adam Mosseri, who heads Instagram, published a long post at the start of 2026 about where the platform is going. His argument was that authenticity is becoming reproducible, that Instagram is working on labelling AI content but will not be able to label everything, and that the platform is shifting attention towards account level signals like consistency and history rather than judging individual posts.
On the search side, Google's published guidance is that it evaluates content on quality rather than on how it was produced, using its experience, expertise, authoritativeness and trustworthiness framework. What it does act on is scaled content abuse, meaning mass produced material with little original value.
Both positions point the same way. Neither is banning the tool. Both are paying attention to whether there is anything of substance behind it.
So how do you decide
The research does not give a yes or no answer, because the right answer changes depending on what the image is doing.
Three questions that seem to hold up against the evidence:
1. Is this image making a claim? If it shows your product, your venue, your results or your work, the trust research suggests a real photo does more for you, and the advertising rules make accuracy a requirement rather than a preference.
If it is decorative, a background, a quote card, a bit of visual furniture, the risk is much lower.
2. Would I be comfortable if someone knew? The transparency research is consistent. Getty's study found 98% of consumers say authentic imagery matters for trust and around 90% want to know when AI has been used. If the honest answer is you would rather people did not find out, that is worth sitting with before you post.
3. Who is my audience? If your customers are mostly under 25, most of them think they can spot it. If they are mostly over 65, most of them think they cannot. Neither fact makes the decision for you, but it should inform it.
Where I have landed
I use AI in my own business, and I also shoot photography, so I have a foot in both camps.
What the research suggests to me is that this is not a moral question about whether AI is lazy. It is a practical question about what each individual image needs to do.
Use it for the jobs where speed matters and nothing is being claimed. Use real images where trust is the point.
And whichever you choose, the thing that decides whether a post works is still what it says and who it is for. A stunning graphic on a post with no clear message will do the same amount of nothing as a mediocre one.
Want to get clear on what you are actually saying before you worry about how it looks?
Content and Cocktails is a half day workshop in Chatham, Kent on 29 September 2026, covering strategy, messaging and audience. Small group, real conversations, and a drink in your hand.
Sources
Include these as links if you want the credibility, or trim to the three or four you lean on most.
British Chambers of Commerce / Atos, Future of Work: AI in the Workplace, March 2026 (UK SME adoption at 54%)
DSIT AI Adoption Research, published January 2026 (marketing as top use case, productivity gains, revenue figures)
Federation of Small Businesses (skills barrier)
Canva Marketing AI Report 2026, surveying 1,415 marketing leaders and 3,547 consumers across seven countries including the UK
The Harris Poll UK, "The Authenticity Trade-Off", April 2026
Klaviyo and Datalily, 2026 AI Consumer Trends, surveying 8,000 consumers, fieldwork December 2025
Gartner consumer survey 2026
Getty Images, "Building Trust in the Age of AI", VisualGPS
Adam Mosseri, Instagram 2026 outlook post, January 2026
Google Search Central, guidance on AI-generated content
ASA and CAP Code, UK advertising standards
