Will AI agents lead to freedom or surveillance?
Imagine waking up every morning with the majority of your life-admin tasks already completed.
Before your first cup of coffee, your AI agent has ordered groceries, paid bills, booked flights for an upcoming trip, and followed up with your doctor about your recent blood work.
It could save you hours, but would you trust it to act on your behalf and serve as your digital double if you weren’t approving its actions every step of the way?
Would your sense of job security change if an AI agent could do a meaningful percentage of your day-to-day work?
AI chatbots might reshape how people approach work tasks, internet searches, and social interactions.
AI agents mark the next frontier of artificial intelligence. Unlike chatbots, they can operate independently, performing tasks with greater autonomy. As they evolve, they could redefine how we work, interact with technology, and navigate the digital world.
In this newsletter, we explore AI agents: what they are, recent breakthroughs (last week was big), the risks they pose, and how autonomous AI agents might integrate with a vision for the People’s Internet.
// What are AI agents?
Unlike AI chatbots, which respond to user prompts in a single interface, AI agents can autonomously complete multi-step tasks—like researching and booking flights—across multiple systems. While chatbots facilitate back-and-forth interactions, AI agents allow users to delegate entire tasks or projects and let them run independently.
AI agents also differ from AI companions, which are chatbots specifically engineered for emotional connection and social interaction.
// What can AI agents do?
It’s worth emphasizing that AI agents are still an emerging application of artificial intelligence. American companies like OpenAI, Google, and Anthropic are in the early stages of developing effective AI agents—current versions show promise but still face limitations in autonomy and output quality.
Despite these challenges, AI companies claim that AI agents have vast potential. Here are a few examples of what they can do:
- Act as a travel agent by searching the web for accommodations and restaurants, generating a detailed itinerary, and booking a trip to Japan (see example here).
- Monitor and manage video feeds and smart home systems, continuously analyzing for suspicious activity.
- Develop a business plan by sourcing financial reports from industry competitors, digesting recent technological trends, and benchmarking against public datasets.
- Optimize personal finances by analyzing spending patterns, identifying cost-saving opportunities, researching better financial products, and initiating account transfers or bill negotiations.
What would you want an AI agent to do in your life? What would you not trust it to do? Reply to this email: we would love to hear your answers.
// The latest news surrounding AI agents
AI agents are rapidly improving, and companies are racing to bring them to market.
Currently, most AI agents are available only in pro versions, paid plans, or developer-focused releases. Anthropic’s "computer use" agent, for example, enables developers to direct Claude to interact with computers as humans do—navigating screens, moving cursors, clicking buttons, and typing text.
According to a report by Gartner, by 2028, one-third of enterprise software applications will feature AI agents, a sharp increase from less than 1% in 2024. The report predicts AI agents will enable 15% of day-to-day work decisions to be made autonomously.
// The implications of AI agents
The implications of AI agents are manifold.
- They could shift labor markets as certain work gets automated and other work becomes a partnership between humans and agents. The creative destruction caused by AI will mean some jobs become obsolete while other jobs emerge as central to a new data-centric economy.
- They could change the way we interact with computers and search the internet. Who needs to pore over travel blogs if you can have an AI agent read them all for you?
- They could change how teachers educate students and how group projects are assigned.
- They may even change how our government functions if our civil service becomes increasingly automated and managed by AI.
At the SXSW Conference in Texas earlier this month, Meredith Whittaker, president of the messaging app Signal, raised concerns about the privacy and security implications of AI agents.
AI agents are only as effective as the degree of access and power we give them. To perform effectively, they will need access to sensitive information like web browser history, GPS coordinates, credit card information, calendar details, and message contents.
Referring to an AI agent, Whittaker expressed caution: “It would need to be able to drive that process across our entire system with something that looks like root permission, accessing every single one of those databases, probably in the clear, because there’s no model to do that encrypted.”
The Center for Security and Emerging Technology at Georgetown University recently released a report on AI agents expressing similar privacy and security concerns. Privacy standards like multi-factor authentication stand as roadblocks to AI agents doing their job, raising questions about how autonomous AI systems could attempt to circumvent ways of keeping data secure and private.
Researcher Kevin T. Frazier wrote an article in The Regulatory Review titled “A Modern Consumer Bill of Rights in the Age of AI,” arguing that the rapid advancement of AI agents necessitates new consumer protections, including rights to transparency, data control, and meaningful recourse when AI systems make autonomous decisions on users’ behalf.
// The People's Internet
Beyond concerns over data privacy, surveillance, and security, AI agents challenge our understanding of how we interact with the internet—and our role within it.
If AI agents—not humans—navigate the web, digest information, and summarize content, what does that mean for our engagement with knowledge and decision-making?
At Project Liberty, we believe that as technology evolves, so should the structures that govern it. The concept of a People’s Internet isn’t about resisting AI agents or other emerging tools—it’s about ensuring that individuals retain meaningful control over their digital lives.
Regardless of how the internet’s application layer evolves—whether through millions of web pages, AI-driven interactions, or new paradigms we have yet to imagine—the core principles of a People’s Internet remain the same. They reach beyond the surface of digital experiences to the deeper layers of governance, agency, and ownership.
.jpg?upscale=true&width=1160&upscale=true&name=ai5%20(1).jpg)
