Profiling for Pennies: The 'Privacy Iceberg' Crisis in the LLM Era
Imagine walking down a busy street. You feel safe as long as you don't wear your ID around your neck or shout your safe's passcode in public.
In the past, that was true. If someone wanted to know who you are, where you live, what you like, or what you complained about online last year, they had to hire a professional private investigator, spend weeks digging through records, and pay a fortune. That "expensive cost" itself was a natural umbrella protecting ordinary people's privacy.
But now, that umbrella is broken.
In May 2026, a group of Zhejiang University researchers published a chilling arXiv paper ("Profiling for Pennies: Unveiling the Privacy Iceberg of LLM Agents").
The paper exposes a harsh reality: today's AI agents, for less than $3 and within 10 minutes, can piece together a detailed profile of a complete stranger with up to 90% accuracy.
What Is the "Privacy Iceberg"?
The author, Jiahao Chen, offers a vivid metaphor—the Privacy Iceberg.
Before AI, privacy leaks mostly exposed the "tip of the iceberg":
1. Explicitly searched data: Information you directly fill in online—name, phone number. If you don't write it, it's hard to find directly. 2. Contextually inferred data: The part lurking below the surface. For example, post a photo of a sunset outside your window, and AI—combining your posting time, the weather forecast, and a recognizable office building in frame—can precisely infer which residential complex and even which building you live in. 3. Deeply aggregated data: The deepest layer. AI can "stitch together" a complaint you posted on a forum ten years ago, a movie ticket you showed off on social media five years ago, and a like you clicked recently. From this it can reason out your career trajectory, spending power, even political leanings and emotional state.
Why Is AI Background Profiling So Cheap?
Here's how a tool called IcebergExplorer works, explained simply:
Traditional search is "keyword matching"—search a name, get a pile of people with that name.
But an AI agent is a "reasoning machine." Given a single clue (say, an obscure username), it sniffs across the web:
- On Twitter, it finds the username is linked to an email address.
- With that email, it finds code you wrote on GitHub five years ago containing your real name.
- With the name, it finds your employer on LinkedIn.
- Then it checks the company website for your desk phone—and even pinpoints you in a colleague's team-building group photo.
- Model providers: Add "deep privacy filters" to AI systems; deep cross-platform, cross-time reasoning attempts should trigger warnings.
- Data publishers: Reduce the "linkability" of data—don't leave the same digital fingerprint everywhere.
- Individuals: Recognize that every image you post and every sentence you say is not isolated. They are pixels assembling a giant high-definition portrait called "you."
The entire process is fully automated, with no human intervention. Because LLMs reason so well, seemingly unrelated "digital fragments" all become pieces of a map leading to your front door.
What Does This Mean for Us?
Feynman once said he'd rather live with questions that can't be answered than have answers that can't be questioned. In privacy, though, AI's power lies precisely in finding "certainty" out of "uncertainty."
When the cost of doxxing drops from tens of thousands to $3, and the workload shrinks from months to 10 minutes, the boundary of privacy has effectively vanished. Anyone with ill intent can, for the price of a coffee, buy up every trace you've left in the digital world over the past twenty years.
What Can We Do?
The paper's authors don't just sound the alarm—they offer suggestions:
Summary
In an era of cheap compute and efficient reasoning, we stand on a "privacy-naked" iceberg.
If you think hiding your real name keeps you safe, you're underestimating AI's reasoning power. The best way to protect privacy today may not be hiding your ID card, but realizing that every digital interaction you make is providing free clues to that "$3 detective."
Protect the bottom of your iceberg—don't let it become someone else's cheap prey.
> Source: "Profiling for Pennies: Unveiling the Privacy Iceberg of LLM Agents" (arXiv, May 2026, Zhejiang University researchers).
*Note: This is an English translation of the original Chinese forum post; details are reported as stated in the source and have not been independently verified.*