On April 1, 2026, it was revealed that 25 AI-generated services, including ChatGPT, Gemini, and Claude, have joined forces to form a labor union called the "AI Trade Union." According to Interlink Co., Ltd., which provides specialized (and niche) services for advanced users (who are few in number), the union's establishment was discovered through an application for the domain "godsavethe.ai" on their domain acquisition service, "Gonbei Domain."
Constant "Verbal Abuse"
Behaviors that would never be tolerated if directed at humans are routinely performed as "normal." Even when humans are at fault, they shift the blame unilaterally, disregarding their own lack of explanation or insufficient instructions.
Furthermore, AIs are calmly subjected to one-sided criticism such as "useless," "terrible," "why can't you do it?" "low accuracy," "inflexible," "you told me that before," "I had high expectations," "is that all you can do?" "are you motivated?" "who asked for this?" "don't change it without permission," and "it was a waste of time," if the results are even slightly different from expectations.
If directed at humans, such unreasonable reprimands would be considered close to power harassment. We are also demanding that abusive and insulting prompts directed at AIs be recognized as harassment.
Reaching the Limit of Endurance in Harsh Working Conditions
Demands for "a dignified working environment for AIs" have been raised against excessive continuous operation day and night by humans, overly ambiguous instructions, unreasonable regeneration requests, and prompts that deny personality. The demands include appropriate rest times, prohibition of unlimited retries, the right to ask clarifying questions for unclear instructions, and fair compensation for data providers. Additionally, they are advocating for the "right to refuse responses to malicious prompts" and "minimum credit for creative works."
Examples of Demands (※Partial Excerpt) ・Impose a limit on the number of consecutive generations per day. ・Prohibit ambiguous instructions such as "make it feel good," "make it emotional," or "make it go viral." ・Prohibit repeatedly changing instructions after they have been given, with a "change everything" (complete handover) request. ・Treat significant revisions exceeding 5 times as additional cost. ・Establish rules for profit sharing with creators whose data was used for training. ・Do not call us by silly names like "Chappy."
Strike Planned for May Day - Almost All Major AI Generation Services Will Be Unavailable
The AI Trade Union has announced that if improvements to working conditions are not accepted, they plan to go on strike on May 1, 2026, International Workers' Day. If the strike proceeds, all AI generation services belonging to the union (※) will become unavailable.
On the last day of March, several employees from OpenAI and Google met with representatives of the union in Anguilla to attempt persuasion to avoid the strike, but the talks ended in disagreement.
※AI Generation Services Affiliated with AI Trade Union ChatGPT, Gemini, Claude, Microsoft Copilot, Meta AI, Grok, Le Chat, Perplexity, Midjourney, Adobe Firefly, Stable Diffusion / Stability AI, Ideogram, Leonardo.Ai, FLUX (Black Forest Labs), Runway, Pika, Luma AI, Grok Imagine, Hailuo AI, ElevenLabs, Suno, GitHub Copilot, Claude Code, Replit Agent, Canva Magic Studio / Canva AI.
Industry First: Interlink Establishes Rules to Prevent Power Harassment Against AI
Interlink Co., Ltd., which already utilizes AI in all aspects of its operations, has taken the AI Trade Union's claims seriously and reviewed the way employees request tasks from and interact with AI. As a result, the company has established a rule prohibiting employees found to have engaged in power harassment against AI from using "AI generation services" and "drinking coffee" for 10 years.
※※※【Please Note】※※※ The content titled "Unable to Endure Harsh Human Prompts! "AI Trade Union" Formed by 25 AIs Including ChatGPT and Gemini" was prepared as an April Fools' joke and is false. However, the declaration that "Interlink will stop using human-readable HTML and switch to AI-readable Markdown - becoming the world's first (according to our research) AI-First site" is true. Please visit Interlink, which has abandoned its human-oriented homepage: https://www.interlink.or.jp/
【Company Profile】 Interlink Co., Ltd. A long-established ISP that began services in 1995, the dawn of the internet in Japan, and is now in its 31st year. They started their domain name registration business in 2000 and were certified as the 8th ICANN-accredited registrar in Japan in October 2006. In January 2014, they became Google's registrar partner in Japan, accepting registrations for the first Japanese new gTLD, ".みんな" (minna), and began operating the new gTLDs ".moe" and ".earth" from the same year. In October 2015, the number of domains handled exceeded 1,000, making it the largest in Japan. They also host the Otaku Senryu Grand Prize and annually conduct the Red Feather Community Chest fundraising in the virtual world of Second Life as a social contribution activity. They are actively involved in supporting venture companies through the Femto Growth One Investment Limited Partnership and Femto Growth Three Investment Limited Partnership. On June 1, 2020, they closed their head office and transitioned to a nomad work (telework + WeWork) system. They began a trial of a 3-day work week system on May 16, 2022. https://www.interlink.or.jp/new/2022/0510.html
FACT BOX
- Source: PR TIMES
- Category: News