Cory Doctorow's Warning: The Spectre of AI Enshittification
In a recent interview with Canadian media outlet BetaKit, author and digital rights activist Cory Doctorow made waves by applying his famous concept of "enshittification" (the planned degradation of technological platforms) to the artificial intelligence market. According to him, the current trajectory of Silicon Valley giants threatens to turn AI into a closed ecosystem designed to capture user data, lock down infrastructure, and extract rents from both businesses and public institutions.
To counter this extreme centralization, Doctorow proposes a radical solution: legally mandating strict interoperability. The activist argues that governments should require open application programming interfaces (APIs) and permit the jailbreaking of closed systems, allowing third parties to graft local and sovereign alternatives onto dominant platforms. This vision challenges the vertical integration model where a single player controls the operating system, cloud infrastructure, and language models all at once.
The Trap of Proprietary Lock-In and Vertical Integration
To understand the scope of this analysis, it is helpful to explain the mechanism of vendor lock-in. When an organization adopts a proprietary AI tool, it is not just choosing a large language model (LLM). It often chains itself to an entire ecosystem: specific data formats, foreign servers, and exclusive APIs. If the vendor decides to raise prices, restrict access to certain features, or unilaterally change its terms of service, the client organization finds itself trapped, as the cost of migrating to another system is prohibitive.
This dependency is heightened by the rise of agentic AI, meaning systems capable of not only answering questions but also executing complex actions autonomously. Without open standards, these agents become exclusive intermediaries between users and their own work tools. As the Competition Bureau of Canada points out in its recent analyses of digital markets, the concentration of technological power harms local innovation and exposes critical infrastructure to single points of failure.
Decoupled Architecture: The ProductivIA Response
The architecture of the Quebec-based ProductivIA platform concretely illustrates how the interoperability principles championed by Cory Doctorow can be translated into code without waiting for regulatory mandates. At the heart of this philosophy is a standardized protocol called assistant_services. This mechanism completely decouples the application layer from the artificial intelligence layer.
In this environment, the core application, the Assistant, orchestrates the platform's other tools using standardized calls. Whether searching for a document, sending an email, or scheduling a task, the Assistant interacts with uniform interfaces. This technical choice offers a major advantage: the administrator of a silo (an organization's sealed logical space) can change the underlying AI model at any time. This makes it possible to switch from a commercial American model to a locally hosted sovereign model, such as Matania, without rewriting a single line of application code.
This open approach is also embodied by the Comparateur IA application. This tool allows users to submit the same query simultaneously to multiple competing models (whether proprietary or open source) to objectively evaluate their performance, bias, and latency. Far from forcing users into a single path, the platform encourages transparent and ongoing evaluation of available technologies.
Data Transparency and Portability
Interoperability is not just about code; it also applies to data. One of the pillars of "enshittification" described by Cory Doctorow is the locking away of user data in unreadable proprietary formats. To counter this trend, ProductivIA's Nuage application offers total transparency over storage.
All files and configurations generated by the applications are accessible in a standardized directory that can be viewed and exported by the user at any time. By guaranteeing native data portability, the platform ensures that the organization retains ownership of its institutional memory, in compliance with the requirements of Quebec's Law 25 on the protection of personal information.
Toward a Resilient Digital Ecosystem
Cory Doctorow's vision reminds us that digital sovereignty cannot simply be decreed: it must be built through architecture. By combining a telemetry-free native operating system like Boreal-OS on hardware, an interoperable application environment like ProductivIA in the browser, and a localized AI engine like Matania, Quebec organizations have a credible alternative to escape tech monopolies. The question is no longer whether we should use artificial intelligence, but how we can integrate it without surrendering our autonomy.