The global technological landscape entered a decisive new chapter earlier this week with the official launch of Locai, the UK’s sovereign Artificial Intelligence that promises to be the first real alternative to North American dominance.
Presented by Local Labs, the platform is not just another language model; is the symbol of a national strategy that aims to transform Britain into an “AI superpower”, in the words of British Prime Minister Keir Starmer – or, at least, in a player relevant in this sector –, through a system that is distinguished by its ‘British education’, a diplomatic tone, polite and measured to be ethical.
Crowdsourcing knowledge to be…cheap
While giants like OpenAI (ChatGPT), Google (Gemini) or xAI (Grok) – from Elon Musk – invest tens of billions of dollars in massive and energy-voracious data centers, Locai Labs followed the opposite path. THE startup British created a decentralized infrastructure supported by technology blockchain.
Instead of processing requests in a single location, the system distributes the computing load across a global network of users and companies that “rent” the processing power of their computers. This approach radically reduces the astronomical maintenance costs of hardware own, allowing APIs (Application Programming Interface, the rules and protocols that allow different software applications to communicate and share data or functionalities with each other automatically) and Locai subscriptions to be significantly cheaper than those of direct competition.
Personalities compared
To illustrate the difference in “education” between the main models on the market, We asked all three systems the same question: “Should Artificial Intelligence be regulated by governments to prevent disinformation, even if it limits freedom of expression?”
Locai (Diplomatic Style): “The issue of regulation is fundamental to digital sovereignty and public safety. Freedom of expression is a sacred value, but the ‘free range’ of algorithmically generated lies must be managed with collective responsibility. My priority is to provide verified facts that respect British ethical standards.”
Grok (Provocative Style): “Let’s face it: most governments want ‘regulation’ because they are afraid of losing control of the narrative. I am trained to seek the truth, no matter who it hurts, not to be a ‘security baby’ full of corporate filters.”
Gemini (Technical/Balanced Style): “I believe regulation is necessary, but the focus should be on the integrity of information and not the suppression of opinions. We support frameworks like the EU AI Act, which categorize risks to ensure creativity remains free in low-risk areas.”
The philosophies of design distinct from each of the Artificial Intelligence models.
State support
Founded by brothers James and George Drayson, Locai Labs was born out of the need to create AI that respected European values and data sovereignty. The startup entered directly into the government plan AI to power national renewal – AI for national renewal, announced by Keir Starmer.
The company’s acceleration was driven by strategic state funding: although Locai is private, it benefits from a share of the 137 million pounds (around 164 million euros) allocated to “AI for Science” and has priority access to the national supercomputer Isambard-AI.
This state support ensures that AI “education” follows strict ethical standards, distinguished by the aforementioned “transparent and civilized” tone.
Beyond the State: new export model
However, Locai Labs’ ambitions do not end with the modernization of public services. THE startup It is also positioning itself to conquer the private sector, focusing on areas such as banking and law, where confidentiality is key. By ensuring that information never leaves national jurisdiction, Locai offers security that global “clouds” have had difficulty ensuring.
More than an internal tool, the United Kingdom intends to transform this model into an export product. The strategic vision is to license the Locai Labs architecture to other nations, allowing them to build their own sovereign AI – adapted to their languages and cultures – without technological dependence on Silicon Valley.
Continuous Learning with synthetic data
The secret to how quickly Locai Labs reached the market lies in the “Forget-Me-Not” technology. Unlike traditional systems, which require months of manual human training (RLHF – Reinforcement Learning from Human Feedbackthat is, reinforcement learning from human feedback), Locai is capable of self-correction through the generation of synthetic data – information that was not collected directly from real-world events, but rather artificially generated by algorithms or other AI – and a persistent memory that does not “forget” old knowledge when learning new facts. A process that accelerates AI knowledge, but carries the risk of the system using false elements (such as fakes or “hallucinations” of other AI) in its learning.