{"id":35,"date":"2026-09-01T10:57:30","date_gmt":"2026-09-01T09:57:30","guid":{"rendered":"https:\/\/blogs.kent.ac.uk\/aiteam\/?p=35"},"modified":"2026-09-01T13:52:47","modified_gmt":"2026-09-01T12:52:47","slug":"ai-generated-content-trust-and-authorship-where-do-we-draw-the-line","status":"publish","type":"post","link":"https:\/\/blogs.kent.ac.uk\/aiteam\/2026\/09\/01\/ai-generated-content-trust-and-authorship-where-do-we-draw-the-line\/","title":{"rendered":"AI-generated content, trust and authorship: where do we draw the line?"},"content":{"rendered":"

A colleague sent me\u00a0some technical\u00a0documentation. It was AI-generated, checked,\u00a0professional\u00a0and useful. Why would I worry?<\/span>\u00a0<\/span><\/p>\n

A colleague published a blog article.\u00a0It\u2019s\u00a0well-written and\u00a0coherent and\u00a0expresses important and useful thoughts. I can\u00a0tell\u00a0it\u2019s\u00a0AI-written. Should I worry?<\/span>\u00a0<\/span><\/p>\n

Online, I\u00a0see\u00a0a fun video of cute kittens squeezing into small spaces. I am not 100% sure if this is AI-generated or not. Should I worry?<\/span>\u00a0<\/span><\/p>\n

The EU has recently brought in the\u00a0<\/span>Code of Practice on Transparency of AI-Generated Content<\/span><\/a>, which\u00a0includes transparency requirements to mark\u00a0some\u00a0content that is generated by AI.<\/span>\u00a0<\/span><\/p>\n

In\u00a0August 2026,\u00a0<\/span>Claude (Anthropic AI)\u00a0announced that it would\u00a0start watermarking<\/span><\/a>\u00a0AI-generated content. But when you look at their announcement in detail,\u00a0it\u2019s\u00a0interesting.\u00a0Anthropic says it will watermark more than the Code of Practice requires, including some categories the Code explicitly exempts.<\/span>\u00a0<\/span><\/p>\n

Does\u00a0a watermark tell us enough about\u00a0the ideas and judgement behind\u00a0a text,\u00a0video\u00a0or image?<\/span>\u00a0<\/span><\/p>\n

It\u2019s\u00a0not a real binary of human-written versus AI-generated\u00a0content. Especially when we write, the line is blurry.\u00a0Even if the watermark for AI can tell us about\u00a0<\/span>provenance\u00a0<\/span><\/b>(was AI involved?),\u00a0it cannot neatly mean that\u00a0<\/span>authorship\u00a0<\/span><\/b>is also by AI, because provenance and authorship are not the same.<\/span>\u00a0<\/span><\/p>\n

What does this mean about trust?<\/h2>\n

\u2460\u00a0Humans want to\u00a0trust humans\u00a0way more\u00a0than AI<\/span>
\n\u2461 Co-creating with AI\u00a0tools could be misunderstood<\/span>\u00a0<\/span><\/p>\n

Take this article. It started out as handwritten scribblings in my paper notebook.<\/span>\u00a0<\/span><\/p>\n

Writing\u00a0screen-free\u00a0considerably helped\u00a0me go slower, read my sources more thoroughly, and develop my thoughts and arguments.\u00a0And writing AI-free helped me\u00a0get a feel for exactly what I want to say and which points are most important to me.<\/span>\u00a0<\/span><\/p>\n

But\u00a0my original paragraphs are annotated, chaotic, all over the place.\u00a0Without\u00a0a\u00a0range of edits,\u00a0restructurings\u00a0and rewrites,\u00a0I\u00a0would not be able to present\u00a0a well-organised\u00a0article\u00a0that\u2019s\u00a0easy to follow and read.<\/span>\u00a0<\/span><\/p>\n

For many people\u00a0(and me), edits are\u00a0a\u00a0tedious, draining\u00a0part of writing.\u00a0The same goes for typing up handwritten notes. As a\u00a0result, I decided to use ChatGPT Edu to help me create something well-written.<\/span>\u00a0<\/span><\/p>\n

You\u00a0know\u00a0who the author\u00a0of this article\u00a0is: me\u00a0(Kerstin Cable from the AI\u00a0team).\u00a0<\/span>
\nBut if you detect an\u00a0\u2018AI\u00a0Tell\u2019 in it,\u00a0you might become unsure if I did write it.<\/span>\u00a0<\/span><\/p>\n

Here is how my writing process went. <\/span><\/h3>\n

I want to share it in detail so you can form your own judgement about\u00a0where and how AI was involved.<\/span>\u00a0<\/span><\/p>\n

    \n
  1. Handwrite out a first draft in my notebook. Research a few developments I reference,\u00a0such as the EU’s Code of Practice on Transparency of AI-Generated Content<\/span>\n

    \"A
    A peek at my original notes<\/figcaption><\/figure><\/li>\n
  2. Take a photo of all my handwritten notes and upload\u00a0them\u00a0to ChatGPT Edu, asking for a conversion of the handwriting to text on the screen. Ask AI to\u00a0maintain\u00a0my exact wording and change nothing in the text.<\/span>
    \n<\/span><\/li>\n
  3. Check whether the AI tool changed anything. Ask for an editable version of the text so I can make corrections where the tool read my handwriting wrong or left a gap marked [unclear]. There is quite a bit of this, because my handwriting isn’t super clear. This is a lot faster than typing it up. In this first pass, I also connect thoughts that had been written out of sequence.<\/span>
    \n<\/span><\/li>\n
  4. Ask for copy-editing in line with the university’s writing voice. I did not let AI make any changes and instead requested approval for each suggestion. This way I\u2019m also teaching AI what\u00a0I’m\u00a0trying to say, catching omissions in the earlier first pass, and revisiting my arguments.<\/span><\/li>\n
  5. Next, an editor’s logic check. I ask it to check my arguments, my\u00a0logic\u00a0and the way I build my argument, like an editor reading something.<\/span>\u00a0<\/span><\/li>\n
  6. Download\u00a0the\u00a0text as a Word\u00a0document\u00a0and work through the points ChatGPT made about where my argument can be strengthened. In the process, I also re-read my text, make my own small edits, and add a new conclusion.<\/span><\/li>\n
  7. Paste the edited version of my article into the tool, asking once again for feedback on my arguments and logic.\u00a0Then edit in the text processor, giving it another read in the process.<\/span><\/li>\n
  8. I\u00a0pick out the occasional sentence I\u00a0don\u2019t\u00a0like yet or find unclear, and test whether ChatGPT Edu can\u00a0suggest\u00a0a clearer way to say it. If I decide the sentence is\u00a0an improvement, I paste it into my article.\u00a0This helps me\u00a0retain\u00a0the original meaning of my statements.<\/span><\/li>\n
  9. Finally, I ask my\u00a0(human)\u00a0colleague to\u00a0read\u00a0the article.\u00a0This is the most intimidating part of the process.<\/span><\/li>\n
  10. I edit once more after the human input, then ask ChatGPT Edu to summarize my article and its main argument. Just to make sure I hit the points I wanted to make.<\/span><\/li>\n<\/ol>\n

    In this process it was important for me to ensure that my words stay my own words, instead of having AI rewrite and rephrase\u00a0everything\u00a0I wrote.\u00a0<\/span>\u00a0<\/span><\/p>\n

    I don’t think that it follows from this process that I am expressing myself in the “very best” way. English\u00a0isn’t\u00a0my native\u00a0language,\u00a0and I have my own cognitive challenges, after all.\u00a0<\/span>\u00a0<\/span><\/p>\n

    But\u00a0these imperfections are a deliberate signal of what makes my thinking and judgement\u00a0human. In a discussion about trust, this felt like\u00a0a good idea\u00a0to me.<\/span>\u00a0<\/span><\/p>\n

    \"Screenshot
    AI doesn’t know what I’m trying to say unless I tell it.<\/figcaption><\/figure>\n

    How does writing happen?<\/h3>\n

    I\u00a0don\u2019t\u00a0always start thinking by hand in a paper notebook.<\/span>\u00a0<\/span><\/p>\n

    Sometimes I might dictate, think\u00a0out loud, talk to\u00a0colleagues\u00a0or even go back and forth in a conversation with ChatGPT Edu.<\/span>\u00a0<\/span><\/p>\n

    People develop thoughts and language in all kinds of ways, and then we translate it into a \u2018dialect\u2019 that is\u00a0appropriate for\u00a0the place we want to share it.<\/span>\u00a0<\/span><\/p>\n

    Each\u00a0university assignment, formal report, social media post,\u00a0announcement\u00a0and\u00a0poster\u00a0has its own conventions for the language and format to use.<\/span>\u00a0<\/span><\/p>\n

    Everything\u00a0that\u2019s\u00a0on our minds \u2014 all that brilliance, beauty,\u00a0erroneous,\u00a0right\u00a0or wrong thinking \u2014 gets translated. And that translation function is different to \u2018generating\u2019 content with AI.<\/span>\u00a0<\/span><\/p>\n

    However, the line again blurs where AI tools introduce\u00a0new ideas,\u00a0claims and language that\u00a0didn\u2019t\u00a0originate with the author.<\/span>\u00a0<\/span><\/p>\n

    Who is the\u00a0author,\u00a0who is\u00a0responsible?<\/h2>\n

    Now, what the EU,\u00a0this blog\u00a0post\u00a0and\u00a0even\u00a0Claude\u2019s\u00a0new watermarking process\u00a0all come back to is this:<\/span>\u00a0<\/span><\/p>\n

    A human\u00a0<\/span>HAS TO<\/span><\/b>\u00a0be\u00a0responsible for\u00a0the quality and truth of AI-generated content\u00a0if we are expected to judge that something we read is trustworthy.<\/span>\u00a0<\/span><\/p>\n

    It\u2019s\u00a0not enough to generate an output, glance at\u00a0it\u00a0and\u00a0publish.<\/span>\u00a0<\/span><\/p>\n

    If I have AI generate an article about this and confidently read it out on camera, you\u00a0may\u00a0trust it more than if I just leave an\u00a0em\u00a0dash in\u00a0(<\/span>em dashes are considered an \u2018AI tell\u2019<\/span><\/a>).\u00a0That is clearly nonsense.<\/span>\u00a0<\/span><\/p>\n

    We\u00a0don\u2019t\u00a0automatically distrust a politician because\u00a0they\u2019re\u00a0reading an elegant speech they\u00a0didn\u2019t\u00a0write.<\/span>\u00a0<\/span><\/p>\n

    The question is not about whether someone used AI or \u2018did their own work\u2019.\u00a0This is often\u00a0seen as a binary division\u00a0when it\u00a0doesn\u2019t\u00a0have to\u00a0be.\u00a0<\/span>\u00a0<\/span><\/p><\/blockquote>\n

    Copy edits, idea exchanges, developing our writing: these are all valuable activities and can\u00a0benefit\u00a0from AI\u00a0assistance.<\/span>\u00a0<\/span><\/p>\n

    The line, however, is one we humans\u00a0haven\u2019t\u00a0learned to \u2018feel into\u2019 yet, because AI is still so new as a writing and thinking tool: when has human judgement been superseded by AI?<\/span>\u00a0<\/span><\/p>\n

    That leads to a more uncomfortable question: am I reading something false or fabricated?<\/span>\u00a0<\/span><\/p>\n

    I\u2019d\u00a0argue that the significant issue\u00a0isn\u2019t\u00a0AI slop, the existence of AI or even how the technology works.\u00a0It\u2019s\u00a0the trust signal.<\/span>\u00a0<\/span><\/p>\n

    A few\u00a0AI\u00a0tells, like\u00a0em\u00a0dashes or a pseudo-lyrical collection of short sentences with rhythmical phrasing, can now\u00a0be enough to cause a reader to distrust\u00a0any text,\u00a0even\u00a0if\u00a0it\u2019s\u00a0been\u00a0well\u00a0researched\u00a0and\u00a0thoroughly thought\u00a0out.<\/span>\u00a0<\/span><\/p>\n

    AI sends a different\u00a0trust signal<\/h3>\n

    When a human writes, we know how to read it critically.\u00a0We\u00a0know that humans can lie, mislead, and repeat errors, and\u00a0that\u00a0it can be done inadvertently or even purposefully\u00a0in order to\u00a0deceive.<\/span>\u00a0<\/span><\/p>\n

    When AI does, we have no clue.\u00a0<\/span>Generative AI is a\u00a0software tool\u00a0much more than a conscious\u00a0author and\u00a0approximates repetitions of what it has seen before rather than judging independently if a fact or sentence is wrong or right.<\/span>\u00a0<\/span><\/p>\n

    Finding your line<\/h2>\n

    We find ourselves in a fascinating, slightly scary moment of human interaction with a new tool.<\/span>\u00a0<\/span><\/p>\n

    It\u2019s\u00a0essential\u00a0to make yourself aware that AI tools cannot judge the truth or accuracy of any person\u2019s claims. They may well reproduce\u00a0accurate\u00a0facts, but they\u00a0don\u2019t\u00a0know if those are facts and truths.<\/span>\u00a0<\/span><\/p>\n

    Institutions and AI companies are beginning to grapple with this new challenge to human trust,\u00a0possibly with\u00a0a few blunt tools that may not get the balance of trust right\u00a0immediately.<\/span>\u00a0<\/span><\/p>\n

    The line is blurry, and the best thing for you as the reader is to stay critically engaged, educate yourself about how AI works and every now and then get the paper notebook out and write and think 100% offline.<\/span>\u00a0<\/span><\/p>\n

    Where is your line?<\/span><\/p>\n","protected":false},"excerpt":{"rendered":"

    A colleague sent me\u00a0some technical\u00a0documentation. It was AI-generated, checked,\u00a0professional\u00a0and useful. Why would I worry?\u00a0 A colleague published a blog article.\u00a0It\u2019s\u00a0well-written and\u00a0coherent and\u00a0expresses important and useful … Read more<\/a><\/p>\n","protected":false},"author":86939,"featured_media":39,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":[],"categories":[315096],"tags":[],"_links":{"self":[{"href":"https:\/\/blogs.kent.ac.uk\/aiteam\/wp-json\/wp\/v2\/posts\/35"}],"collection":[{"href":"https:\/\/blogs.kent.ac.uk\/aiteam\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/blogs.kent.ac.uk\/aiteam\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/blogs.kent.ac.uk\/aiteam\/wp-json\/wp\/v2\/users\/86939"}],"replies":[{"embeddable":true,"href":"https:\/\/blogs.kent.ac.uk\/aiteam\/wp-json\/wp\/v2\/comments?post=35"}],"version-history":[{"count":5,"href":"https:\/\/blogs.kent.ac.uk\/aiteam\/wp-json\/wp\/v2\/posts\/35\/revisions"}],"predecessor-version":[{"id":44,"href":"https:\/\/blogs.kent.ac.uk\/aiteam\/wp-json\/wp\/v2\/posts\/35\/revisions\/44"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/blogs.kent.ac.uk\/aiteam\/wp-json\/wp\/v2\/media\/39"}],"wp:attachment":[{"href":"https:\/\/blogs.kent.ac.uk\/aiteam\/wp-json\/wp\/v2\/media?parent=35"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blogs.kent.ac.uk\/aiteam\/wp-json\/wp\/v2\/categories?post=35"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blogs.kent.ac.uk\/aiteam\/wp-json\/wp\/v2\/tags?post=35"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}