How decentralised platforms can restore control in a world of generative overreach

In the rapidly evolving AI economy, power is consolidating and so are the lawsuits. Microsoft is now the latest tech giant facing legal action over alleged copyright violations in training its AI models. According to Coingeek, the company is being sued by a group of prominent authors who claim that more than 200,000 pirated books were used without consent to train Microsoft’s generative models. The plaintiffs are demanding up to $150,000 per misused work and a permanent injunction to stop such practices entirely.

Microsoft, the parent company of OpenAI investor and partner in various LLM initiatives, is not alone in this spotlight. Other tech conglomerates such as Meta, OpenAI, and Alphabet (Google’s parent company) are facing similar scrutiny for their opaque data practices. While some courts have ruled in favour of the tech companies, citing “fair use” or lack of substantial similarity, these outcomes often hinge on legal technicalities rather than ethical clarity.

The ethics of data extraction

Large language models (LLMs) rely on massive volumes of training data. While some of this data is publicly available, the most valuable training sets often include copyrighted works such as books, articles, and media, scraped without consent. The result is a growing disconnect between those who create knowledge and those who monetise it.

As the recent filings against Microsoft state: “Microsoft created a computer model that is not only built on the work of thousands of creators and authors but also built to generate a wide range of expression that mimics the syntax, voice, and themes of the copyrighted works on which it was trained.” (Coingeek, 2024)

Authors, artists, and publishers are now turning to tools like data poisoning and blockchain-based protection layers to safeguard their work. These mechanisms aim to make datasets unusable for unauthorised training or to trace copyright breaches across the data lifecycle. But fundamentally, these are reactive solutions in a system built without consent at its core.

The hype and hurdles of agentic AI

While Microsoft and others defend their models in court, many of these same AI projects face commercial uncertainty. A recent report by Gartner predicts that 40 percent of all agentic AI initiatives those using autonomous AI agents capable of making decisions will be abandoned before 2027. The reasons include rising costs, unclear business value, and inflated expectations.

Agentic AI is being positioned as the next frontier, with companies like Salesforce and Oracle investing heavily in experimentation. Yet Gartner notes that only 130 of the thousands of so-called AI agents on the market are authentic. Many are simply rebranded chatbots.

For companies and users alike, the result is a confusing landscape. One where proprietary models are trained on murky data, platforms are built around surveillance capitalism, and enterprise promises outpace practical delivery.

The alternative: decentralised, ethical, human-centric

This is where platforms like Unyted come in. Built not to extract value from users, but to create value with them. At Unyted, we believe that collaboration and innovation should not come at the cost of privacy, consent, or clarity. We do not scrape, monitor, or monetise user behaviour. Our decentralised infrastructure means that control does not rest in a single corporate server but in the hands of the people who use it.

In contrast to the proprietary, closed-source systems currently under fire, Unyted was born from the mission to create a secure and compliant platform for people and organisations. We are at the forefront of ethical, decentralised technology championing user-first design, transparent data handling, and collaborative innovation.If you want to see what secure, compliant and decentralised collaboration looks like in action, book a free demo with Unyted today.mo with Unyted today.ed today.