AI-generated content will not disappear even with watermarking and labelling tools, according to Alex Cui, co-founder and chief technology officer of AI detection platform GPTZero.
Writing in City AM, Cui said regulation alone is not a 'silver bullet' and that watermarking tools are 'premature and prone to manipulation'. He cited recent examples of AI misuse, including a Harvard professor's AI-assisted column published by the Financial Times and research reports from Big Four firms such as PwC and KPMG containing AI hallucinations.
Cui noted that GPTZero research suggests roughly 44 per cent of LinkedIn content is currently AI-generated, and without intervention it could reach 100 per cent by 2031. He pointed to LinkedIn's 'Seems like AI slop' reporting button introduced in July and Anthropic's watermarking tools for Claude-generated text, which align with the EU AI Act's Transparency Code.
He argued that AI should be used to organise complex information, not as a substitute for human judgement and creativity. 'People care about what people think, not machines,' he wrote.