The issue of trust was always going to come up eventually when discussing AI. After all, would you really trust a robot with brain surgery or to fly a plane? Maybe in 50 years or so when we’ve all become more comfortable with the idea of automation. For now, however, in the fledgling days of the technology, we’re all a little understandably uncomfortable with the idea of AI taking the reins entirely. That includes the content we consume.

In a world powered by AI, literally anyone can generate a polished article, ad or news piece in the time it takes to boil a kettle. But these digital wunderkinds are far from infallible.

Indeed, it’s been found that these models often “hallucinate” content by inventing facts or misquoting sources because they stitch together patterns in data without real-world verification. In one BBC study, the AI chatbots gave respondents completely inaccurate news (for example, stating that Rishi Sunak was still prime minister). 

AI can generate content quickly and fluently and it can be incredibly helpful, but it can also twist or fabricate information. As a result, even seemingly authoritative-sounding AI content needs to be treated with caution.

News publishers themselves are experimenting with AI, often for routine pieces. In the UK, for example, the publisher of the Mirror and Express has admitted using ChatGPT to help reporters write local weather and traffic updates. But early trials highlight the risks: one technology news site found that AI-generated financial reports contained multiple factual errors. 

These mistakes echo dire warnings from industry leaders. The BBC’s news chief Deborah Turness cautioned that “Gen AI tools are playing with fire” and could undermine the public’s “fragile faith in facts”. In other words, AI is a powerful writing tool, but it cannot replace human judgement, and readers should not assume its output is automatically accurate. 

In other words, don’t trust it without doing your own research and backing it up. But how do we know what to trust when we don’t know what is and what isn’t AI?

Public Scepticism

Surveys show readers in both the UK and the US are wary of AI-written news. A Reuters Institute report found that a clear majority of people would be uneasy if most news was produced by AI. 63% of UK respondents and 52% of US respondents said they would be uncomfortable with that scenario. 

In practice, audiences trust human oversight far more. A U.S. survey by Rutgers University found 62% of Americans “somewhat” or “a lot” trust content by mainstream journalists, whereas only 48% trusted content known to be AI-generated. 

In the UK, a YouGov poll similarly showed that Britons give much more credibility to articles involving humans: about 48% said they would trust an article written and edited by people, but only 13% would trust one written and edited by AI. In fact, 72% of Britons said they would distrust an article if both the reporter and editor were AI.

This widespread scepticism has practical consequences. One media survey in the U.S. found that reader trust nearly halved whenever people thought an article was AI-generated – even if it wasn’t. Advertisers notice this too: brands see lower purchase intent when their ads appear next to content perceived as “AI-made”.

Both UK and US audiences tend to trust human journalism more for obvious reasons. Human beings have lived experience and their own opinions and rationale to work with. This is one area in which AI is still scrambling in the dark. 

As such, unsurprisingly, many in the UK now expect transparency. A recent YouGov study found 79% of Britons want news outlets to clearly label any AI assistance in an article. (BuzzFeed, for example, now marks AI-written travel pieces as such and rightly so, even though these articles could be considered “puff pieces”) This suggests the public wants to know who made the content before deciding whether to believe it.

How to spot reliable content

Even as AI floods the web with new text, traditional media literacy still applies. Users are advised to verify claims before trusting them. In practice, this means cross-checking suspicious facts or quotes with reputable sources. 

For instance, one expert guide advises that readers should “always cross-check AI-generated facts with authoritative sources” like academic papers, government websites or established news outlets. 

If an AI-written article cites data or quotes, search for those details on trusted sites (for example, a major news site or an official press release) to confirm they really exist. Look at the byline or publisher: content from well-known newspapers or broadcasters (which have professional editors) is more likely to have been vetted.

• Check multiple reliable sources: If an unusual claim appears, see if any major news organizations or official sources report the same thing. Remember that if only obscure blogs or social media say it, it may be false.
• Use fact-checkers and AI tools: Organizations like Snopes, FactCheck.org or Full Fact can verify viral claims. In fact, fact-checkers themselves are using AI: the UK charity Full Fact has built an AI system to flag misleading claims in news articles. Google and social media platforms are also rolling out AI-powered fact-check assistants.
• Examine the content style: AI text can sometimes be a bit repetitive, overly formal or generic. If a piece lacks specific details or sounds strangely uniform, that’s a red flag.
• Beware of sensationalism: If a story makes an outrageous claim with no obvious evidence or citations, it’s wise to be cautious until it’s confirmed. Misinformation often tricks readers with clickbait headlines.
• Stay up-to-date: AI systems train on data up to a point and may be unaware of recent events. Always check the date. If an AI blurb claims recent events that seem off (like outdated leadership info), verify with an up-to-date source.

Following these habits (think of them as digital fact-checking “lateral reading” steps) should help sift trustworthy content from fiction nine times out of ten. In short, treat AI-generated text the way you’d treat any news: by confirming facts independently and considering the credibility of the publisher.

Another Opinion

“Trust is an interesting word – I don’t trust or mistrust AI more than I trust or mistrust a human in a marketing team. I believe fundamentally it’s about trusting brands and publishers around what they do and don’t let onto their channels. There will be bad actors, click baiters and straight scammers using AI for all the wrong reasons – and I do believe there needs to be some sort of ‘fact-checking’ service just to ensure that complete misinformation isn’t shared. AI is just a tool, but I do believe our senses need to be heightened and we all need to be questioning what we see before jumping to conclusions – is this real? Is this content there to serve a genuine purpose?”
— Alex Bodini, CEO at Spin

Bodini’s view reinforces a key point: AI is neither inherently trustworthy nor evil – it’s a tool. What matters is who is using it and why. Whether content is human-written or machine-generated, readers should stay alert and ask the same questions: Does this source have a track record of accuracy? Has anyone verified these facts?

In the end, trust comes down to transparent editorial standards and a questioning mindset. As all the experts are advising, “stay critical, ask questions, and confirm facts” before accepting or sharing content whether it’s AI-made or not.