Public Libraries Become Ground Zero for the Opt-Out Movement

The Workshop That Broke the Internet

When Hannah Cyrus, an information specialist at Bangor Public Library in Maine, published a journal article describing a workshop she had developed, she expected the modest circulation typical of library science literature. Instead, her inbox exploded. Librarians from across three continents reached out, asking for slides, handouts, and permission to replicate the program in their own communities.

The workshop's title was simple: Avoiding AI. Its premise was even simpler - show people, step by step, how to turn off the generative features that have proliferated across consumer devices and platforms over the past eighteen months.

Cyrus typically draws a dozen attendees to her digital literacy classes. For the first Avoiding AI session, she capped registration at thirty, opened a waitlist, and added a Zoom livestream. Seventy people attended. The second session drew identical numbers.

At DailyTechWire, we've tracked the rollout of consumer AI features across major platforms since late 2024. What distinguishes this moment from earlier waves of feature creep is the sheer velocity of deployment and the scarcity of opt-out paths. Apple Intelligence, Gemini integration in Android, Microsoft Copilot embedded in Windows, generative search summaries in Google - each arrived with minimal user consultation and, in many cases, opt-out controls buried three menus deep.

From Maine to Philadelphia, a Pattern Emerges

Charlie Bailey, a librarian in South Philadelphia, came across Cyrus' work and decided to adapt the workshop for his own branch. He anticipated interest. He did not anticipate the Instagram post announcing the event would pull over two thousand likes and more than two hundred shares - orders of magnitude beyond the library's typical social engagement. He scheduled a second session before the first had even taken place.

The workshop Bailey runs unfolds in a children's classroom, alphabet rug underfoot, but the twenty adults in attendance are there for a different kind of literacy. Bailey opens with a primer on how large language models and consumer generative tools function - not a technical deep dive, but enough context for participants to understand what happens when they interact with these systems. Then he walks through the major platforms one by one: iOS, Android, Gmail, Microsoft Office, Google Search. For each, he projects step-by-step instructions on how to locate and disable specific features.

The format is instructional, but the atmosphere is communal. Attendees trade tips. One person shares that appending a specific URL parameter to Google searches will suppress AI-generated summaries. Bailey writes the string on a whiteboard, next to the login credentials for the library's teen Wi-Fi network.

The Straw That Broke the Camel's Back

The questions Cyrus fielded at her reference desk over the past year tell the story. Why is this tool trying to write my email for me? Why am I getting a summary of a one-sentence message I can already read? The frustration wasn't rooted in technophobia. It was rooted in the erosion of choice.

Bailey frames the work as an extension of digital literacy - a core mandate for public libraries in an era when access to technology is no longer optional. But he also sees it as something more fundamental: helping people reclaim autonomy over tools that increasingly shape daily life without their consent.

One workshop attendee, Gabrielle, articulated the tension many in the room felt. She described the environmental cost of generative AI - the energy-intensive data centers, the water consumption, the carbon footprint of inference at scale - and the dissonance of having these systems pushed into her workflows at work. At the same time, she acknowledged the potential of machine learning in domains like medical research. The issue wasn't the technology itself. It was the lack of agency in how and when it entered her life.

Another participant, Johnny, raised a different concern: the infrastructure buildout. He pointed to the proliferation of data center construction across suburban and exurban areas, often adjacent to residential neighborhoods, and the reality that homeowners have little recourse once these facilities break ground.

What Opt-Out Really Means

Neither Cyrus nor Bailey position themselves as anti-technology. Cyrus uses optical character recognition daily in her work digitizing archival materials. Bailey teaches coding workshops for teens. Both recognize that the algorithms powering generative AI are part of a broader family of machine learning techniques that have been embedded in computing for decades.

What they're pushing back against is forced adoption - the design pattern in which new features default to on, opt-out mechanisms are obscured, and user preferences are overridden with each software update. This is not a new dynamic in consumer technology, but the stakes feel different when the features in question ingest user data at scale, generate output that can be misleading or fabricated, and impose environmental costs that are largely externalized.

The workshops don't advocate for rejecting AI wholesale. They advocate for informed choice. They teach people how to read privacy policies, how to interpret data-sharing settings, how to recognize when a tool is operating in the background without explicit invocation. They treat technology as something users should control, not something that controls users.

A Movement Without a Manifesto

Cyrus described the response to her work as unprecedented in her career. Librarians don't typically go viral. But the emails kept coming - from the UK, from Australia, from Canada, from small-town branches and urban systems alike. Each one asked the same question: Can we run this in our community?

There's no central organization coordinating these workshops, no formal curriculum, no branding. What exists is a loose network of information professionals who saw a need, built a response, and shared it freely. The structure is peer-to-peer, the ethos is open-access, and the motivation is rooted in a professional commitment to equitable access to information and tools.

Bailey noted that as someone who works in information services, it felt affirming to see skepticism toward AI. Not cynicism, not reactionary rejection, but thoughtful skepticism - the kind that asks who benefits, who decides, and what gets lost in the transaction.

The scale of interest suggests this sentiment is widely shared. The people filling library classrooms to learn how to disable Gemini or Apple Intelligence are not Luddites. Many work in tech-adjacent fields. They use smartphones, cloud storage, algorithmic recommendation engines. What they're resisting is the collapse of the boundary between tool and mandate.

The Broader Context: Adoption Without Consent

The forced-adoption dynamic Bailey and Cyrus describe is not unique to generative AI, but it has intensified in this cycle. Platform incentives - driven by investor expectations, competitive pressure, and the need to demonstrate AI integration as a value proposition - have compressed the timeline between product development and user-facing deployment. Features that might once have undergone extended beta testing or opt-in pilots are now shipping as defaults.

This creates friction not just for individual users but for institutions. Schools, hospitals, legal practices, and newsrooms are navigating the compliance and liability implications of tools that generate content, make recommendations, or automate decisions without clear lines of accountability. The question of who is responsible when an AI summary is factually incorrect, or when an auto-generated email introduces ambiguity into a professional exchange, remains unresolved in most jurisdictions.

At the same time, the infrastructure required to support consumer-scale generative AI is reshaping physical landscapes. Data centers are among the fastest-growing categories of industrial construction in North America and parts of Asia. These facilities require enormous amounts of electricity and water for cooling, and their placement - often in areas with lower land costs and favorable tax treatment - can strain local grids and water tables. For residents in affected areas, the tradeoff between economic development and environmental impact is not abstract.

What Comes Next

The library workshops are a pressure-release valve, but they're also a signal. They indicate that a meaningful segment of the user base is willing to invest time and effort to push back against default settings, even when doing so requires navigating opaque menus and resetting preferences after every update.

Whether this translates into sustained behavior change or broader policy shifts remains to be seen. Platform designers have decades of experience in dark patterns - interface choices that nudge users toward preferred actions - and the incentive structure heavily favors re-engagement. A user who disables AI features today may find those features re-enabled after a software update tomorrow, with opt-out paths moved or renamed.

But the workshops also represent something harder to quantify: a reassertion of user agency in an environment where agency has been steadily eroded. They treat technology as a domain where literacy matters, where understanding how systems work is a prerequisite for using them on your own terms. And they locate that work in public libraries - institutions that have historically served as counterweights to commercial imperatives, spaces where access is not contingent on monetization.

Bailey and Cyrus both plan to continue running the workshops as long as demand persists. Other librarians are developing their own versions, tailored to local contexts and user needs. There is no unified strategy, no roadmap, no venture funding. Just a growing network of people teaching other people how to turn things off.