Background & Context§
LibreOffice, the open-source office suite maintained by The Document Foundation (TDF), has been a staple alternative to Microsoft Office for nearly two decades. Its latest release, version 26.8, launched on August 26th and quickly became the most downloaded update in the project's history. While the suite has long been praised for its compatibility and zero-cost licensing, the surge in popularity appears tied to a deliberate anti-AI stance. TDF publicly declared that LibreOffice does not include generative AI features, framing this omission as a principled feature rather than a gap. In an era where AI integration is often a selling point, this decision has sparked widespread media coverage and renewed interest in the software.
The News: What Happened Exactly§
Within one week of its release, the LibreOffice 26.8 installer was downloaded more than 1 million times, excluding updates delivered through Linux distribution repositories. This milestone marks a record for the project, which has historically seen steady but modest adoption. The Document Foundation celebrated the achievement, but its subsequent announcement clarified that the download spike was not solely due to technical improvements in writing systems or typography. Instead, TDF attributed much of the attention to its public stance against integrating generative AI by default.
A day after the download record was reported, TDF published a post titled "Yes, no AI is now a feature," authored by Italo Vignoli. The article emphasized that the foundation "does not reject artificial intelligence out of hand," but insists that current AI technologies fail to meet a comprehensive set of principles required for default inclusion. These principles, though not exhaustively listed in the source, center on user privacy, data sovereignty, and ethical deployment. TDF argues that generative AI, as commonly implemented, often requires sending documents to external servers, raising significant privacy and security concerns.
Furthermore, TDF took a jab at competitors who have rushed to embed AI assistants. According to the foundation, such integrations often serve as justification for price increases and encourage users to keep all documents within a proprietary infrastructure. By contrast, LibreOffice remains committed to local processing and user control. For now, TDF recommends community-developed plugins for those who wish to add AI capabilities, allowing users to opt in rather than having AI forced upon them. This approach aligns with the project's long-standing philosophy of transparency and user empowerment.
Historical Parallels & Similar Incidents§
This is not the first time a software project has gained traction by rejecting a prevailing trend. In the early 2000s, Apple's decision to exclude Adobe Flash from the iPhone was initially met with criticism. Steve Jobs famously penned an open letter titled "Thoughts on Flash" in 2010, outlining technical and security reasons for the omission. At the time, Flash was ubiquitous for web video and interactive content. Apple's stance was seen as restrictive, but it ultimately pushed the web toward open standards like HTML5. Similarly, LibreOffice's refusal to embed generative AI may accelerate the adoption of privacy-preserving alternatives and community-driven plugins. Both cases illustrate how a controversial omission can become a defining feature, though Apple's move was strategic and proprietary, while TDF's is rooted in open-source ethics.
Another parallel is the rise of privacy-focused search engine DuckDuckGo. Launched in 2008, it differentiated itself by not tracking users, at a time when Google's data collection was the norm. DuckDuckGo's growth was gradual but steady, and it now handles billions of queries annually. Like TDF, DuckDuckGo did not reject technology outright but prioritized privacy as a core principle. The lesson is that a significant user base values data protection over convenience, especially as AI becomes more pervasive. However, unlike DuckDuckGo, which built its own indexing technology, LibreOffice relies on third-party plugins for AI, which may limit its appeal to users seeking seamless integration. Nevertheless, the download record suggests that a substantial segment of the market rewards principled stands.
The Technical and Strategic Implications§
TDF's decision to forgo default AI features is not merely ideological; it has practical ramifications. Generative AI models, such as large language models (LLMs), typically require substantial computational resources and cloud infrastructure. Integrating them into an office suite would introduce dependencies on external APIs, increase attack surfaces, and complicate offline usage. By avoiding this, LibreOffice maintains its lightweight footprint and compatibility with air-gapped environments. For developers, this means the core codebase remains free of AI-related bloat, but it also opens opportunities for plugin ecosystems. A community plugin could, for instance, use a local LLM via Ollama or llama.cpp to provide writing assistance without sending data to the cloud. Below is a simplified Python snippet illustrating how a local AI plugin might interface with LibreOffice's UNO API:
import uno
from com.sun.star.beans import PropertyValue
def get_document_text():
ctx = uno.getComponentContext()
desktop = ctx.ServiceManager.createInstanceWithContext(
"com.sun.star.frame.Desktop", ctx)
doc = desktop.CurrentComponent
text = doc.getText().getString()
return text
# Example: send to local LLM for summarization
def summarize_with_local_llm(text):
# Assume a local endpoint at http://localhost:11434/api/generate
import requests
response = requests.post('http://localhost:11434/api/generate', json={
'model': 'llama3',
'prompt': f'Summarize: {text}',
'stream': False
})
return response.json()['response']This approach keeps data on the user's machine, aligning with TDF's privacy principles. However, it places the burden of setup on users, which could be a barrier for non-technical audiences. TDF's recommendation of community plugins effectively delegates AI integration to those who truly need it, avoiding the one-size-fits-all model of competitors. It also allows the foundation to sidestep the ethical and legal complexities of training data provenance and model bias, which have plagued other vendors.
In the broader AI landscape, this stance challenges the narrative that every software must incorporate generative AI to remain relevant. It signals that a significant user base—particularly in NGOs, government agencies, and enterprises with strict data governance—prioritizes control over cutting-edge features. As AI regulation evolves, TDF's principled approach could serve as a blueprint for others. Whether this translates into sustained growth remains to be seen, but the record downloads suggest that "no AI" can indeed be a compelling feature.