Almost everyone I know who uses email for work these days, has either made use of ChatGPT or other LLMs as a tool to help with their writing, or they’ve at least experimented before rejecting it.
It has allowed people who may not necessarily have the educational background or literary flair, to express themselves professionally, sounding diplomatic in tricky situations, and speaking with a confidence they might not have had before. Some authors simply use it to check grammar, but for many regular people, the heavy lifting is now done by the algorithm.
AI writing is now so mainstream that jurisdictions can’t move fast enough to limit its use in schools and other fields. It’s shown up in politics too, with AI-generated graphics and memes fuelling social media arguments that genuinely shape public debate, and crucially, elections.
That’s part of why the EU’s AI transparency rules which kicked in from August 2026, requires companies to clearly label AI-generated content, whether that’s a chatbot reply, an image, or a block of text. Anthropic has started coming up with a way to watermark these.
The watermarking involves embedding identifiers directly into the file, so AI-generated content can be told apart. Almost everyone agrees the goal is to have transparency and protection of original works.
An unacknowledged watermark could easily turn into an accusation of plagiarism, in a job application, a university assignment, or a report with your name on it. These announcements are timely, given the recent tragic passing of Jason Arday, who was accused of plagiarism and consequently harassed to the extent that he took his own life. Given a recent study found only 38% of AI image tools currently watermark adequately, the systems meant to protect you aren’t fully reliable yet either.
Be upfront: a quick note like “drafted with AI assistance, edited by me” does you far more good than staying quiet. Save all your prompts and don’t go stripping watermarks or metadata off anything, it’s tempting, but that’s exactly what turns a grey area into a real headache.
These are tools that are now part of our modern (and future) lives. Tech-companies are driven to pass off their AI models as real humans, as much as possible. The line where the uncanny valley ends is getting more and more blurry.
In a society that is increasingly adopting the widespread use of bio-computers, and very convincing humanoid robots, for our own safety and protection, we need to be able to tell the difference, between a human touch, and a computer’s hallucination.
