Joe Tidy, BBC Cyber Security Correspondent
[Getty]
Organizations around the world are competing to develop a universally recognized label to signify products and services that are ‘human-made’ as part of a trend against the rising use of artificial intelligence (AI).
Statements such as ‘Proudly Human’, ‘Human-made’, ‘No A.I’, and ‘AI-free’ are beginning to appear in movies, marketing, books, and websites.
This phenomenon is a response to concerns that AI automation could make jobs or even entire professions obsolete.
According to BBC News, at least eight different initiatives are attempting to launch a label, hoping to gain global recognition like ‘Fair Trade’ for ethically produced goods.
However, with numerous competing labels and the unclear definition of ‘AI-free’, experts say that if a consensus on a single standard cannot be reached, consumers may become confused.
Dr. Amna Khan, a consumer expert from Manchester Metropolitan University, states: ‘AI is causing significant disruption, and the competing definitions of what constitutes ‘human-made’ are confusing consumers.’
She told BBC News: ‘Establishing a universal definition is crucial for building trust, providing clarity, and enhancing confidence.’
Major companies and non-profit organizations in the UK, Australia, and the USA have launched labels and seals. [BBC]
The movement to establish a no-AI certification system has arisen after generative AI tools have been used to replace human labor and creativity in various industries, including fashion, advertising, publishing, customer service, and music.
The organizations attempting to introduce relevant labels include both companies and non-profits located in the UK, Australia, and the USA.
How Certification Works
Some labels, such as ‘no-ai-icon.com’, ‘ai-free.io’, and ‘notbyai.fyi’, can be downloaded for free or for a fee, with little to no review mechanisms.
Other systems, like ‘aifreecert’, require fees and have stringent review processes to verify whether a product has used AI. Reviewers utilize professional analysts and AI detection software.
However, AI experts suggest it will be quite complicated to reach a consensus across industries on what ‘true human creation’ actually means, given that AI has integrated into many everyday tools.
Not By AI offers verification services, and you can pay to download and use their badge. [Not By AI]
AI researcher Sasha Luccioni explains: ‘AI is now ubiquitous and has integrated into various platforms and services, thus making it very complex to define what ‘AI-free’ really means.’
‘From a technical standpoint, this is difficult to achieve. I think AI exists on a spectrum; hence we need a more comprehensive certification system rather than simple AI/no-AI binary classification.’
No Generative AI
Some people believe the line should be drawn based on whether generative AI is used—which refers to chatbots that can generate text, code, music, or videos based on human prompts.
In the credits of the 2024 thriller ‘Heretic’, starring Hugh Grant, the production team included a statement: ‘No generative AI was used in the making of this film.’
The film distributor ‘The Mise en scène Company’ has also adopted this concept, recently placing a ‘No AI Used’ seal on its latest movie poster. This film was primarily written, directed, and edited by the same individual.
The company has also published its classification criteria online, hoping others in the industry can adopt it.
Company CEO Paul Yates states: ‘We support the AI industry and believe it’s an exciting time, but we think that due to the emergence of AI content, human-created content will gain a premium in the market, and we wish to capitalize on that.’
Impacts of AI
The art industry is particularly flooded with products made by AI, thus it seems to be the current focal point of the movement against the use of AI.
Entire books and films can now be produced using AI, often faster and cheaper than traditional methods.
The Bollywood company ‘Itelliflicks’ specializes in making AI films and prides itself on that.
However, sometimes, products reliant on AI do not clearly inform consumers.
Intelliflicks has entirely utilized generative AI tools to produce many movies. [Intelliflicks]
The viral band ‘Velvet Sundown’ from last year was similarly revealed to be entirely generated by AI.
In publishing, large houses like Faber and Faber are beginning to stamp ‘Human-Written’ on some books.
Author Sarah Hall has requested this stamp to be placed on her novel ‘Helm’, characterizing the use of book content in training AI models as ‘massive creative theft’.
Nevertheless, Faber and Faber has not outlined how it defines ‘human-written’ books or what kind of reviews would ensure no use of AI.
The UK company ‘Books by People’ agrees that a reliable standard is essential to regulate how human authorship is disclosed.
Co-founder Esme Dennys states: ‘Publishers face a new environment—books can be completed in minutes instead of months or even years. Readers can no longer tell whether a book reflects human experience or machine imitation.’
The company has collaborated with five publishers and launched its mark on the book ‘Telenova’ published in November last year.
‘Books by People’ issues stamps to publishers who pay for anti-AI audits. [Books by People]
‘Books by People’ charges publishers and requires them to complete a questionnaire detailing their publishing processes and how they verify authorship. The company also conducts regular spot checks on book samples to detect potential AI writing.
In Australia, the competing company ‘Proudly Human’ employs a similar but more stringent system to ensure authors do not use generative AI. Their reviewers check at every stage of the publication, including verifying any alterations between the manuscript and ebook versions.
The company is set to announce partnerships with several major publishers and plans to expand its business into music, photography, film, and animation.
Company head Alan Finkel states that systems like these are crucial as industry efforts to analyze and label AI-generated content have consistently fallen short.
He says: ‘We need certification on ‘human origin’, but self-certification is not enough; hence we established a comprehensive validation process to ensure content is truly human-created.’