SNIFFO
Spot misinformation and AI-generated content before you share it.
Social media is flooded with bots, AI-generated content, and misinformation. Sniffo is an extension that scans posts and accounts as you scroll and flags three things: misinformation, AI-generated content, and account-level risk. Stay informed, stay skeptical.
Team: Silvia Zhang, Keller Wang, Bri Mai
My role: Information architecture for detection results and real-time feedback, risk-information hierarchy and visualization, UX/UI, visual and brand expression, concept-video editing
Deliverables: Prototype, wireframes, data visualization explorations, marketing reel
Two moments, two speeds
Scrolling
Sniffo sits quietly in the dynamic island and scans in the background.
Something is flagged
A small badge appears, here “Mis” for possible misinformation. No pop-up.
Expand
Three dials for MisInfo, AI Gen and Auth Risk, plus one line of context.
Choose
Keep scrolling, or jump to the full report in the app.
Scroll with Confidence
While scrolling, there is no time to investigate. A small signal in the dynamic island is enough to say: look twice, before you believe or share.
Open the report
Misinformation first: fact-check sources and context from similar posts.
AI-generated?
A combination analysis and the likely model signature behind the text.
Who posted it?
Account risk from posting frequency and past reliability.
Saved
Every result is saved in the Sniffo app, ready to revisit.
Dive Deeper with Clarity
When a post still feels off, one tap opens a layered report with misinformation signals, AI-content detection, and the author’s track record.
01
Sniffo scans and alerts during scroll through dynamic island.
02
Sniffo's in-app interactive report on post's misinformation.
03
Sniffo's interactive analysis on post's AI generated content.
04
Sniffo analyzes account risk based on past posting behavior.
Why a glyph?
While exploring, we moved from showing information to delivering a signal. A glyph carries several dimensions of information in a very small space, with very little interference.
Initial tests on using a single glyph to convey risk. We tested arcs, rings, and dots to determine the most effective design for small scales.
Explored multiple shapes, densities, and placements to understand how micro-glyphs communicate risk at a glance.
Three small dials, one line of context. Enough to decide in a glance: keep scrolling, or open the report. The glyph became the bridge between complex data and instant understanding.
The problem
“We accidentally digest and spread misinformation more often than we realize on social media.”
Why news feeds like X? On entertainment feeds like Instagram, a fake post matters less. On a news feed, an AI-generated post is more dangerous, because it can pass as the real thing.
Source: Pew Research Center, Oct 2025 · 5,153 U.S. adults. About news in general, not only AI content.
Limited time
On a phone, people rarely stop to research a post. The check has to cost almost nothing.
Pulled out of flow
Switching between platforms and headspaces is unintuitive, and it breaks focus.
No instinct for AI content
Most everyday people in the US are not well equipped to spot misinformation or AI on social media.
Design decisions
How can detection fit into the way people already scroll, instead of interrupting it?
Put the signals in the dynamic island, a system surface people already glance at, so there is nothing new to learn.
How can a technical, cold tool feel like something an everyday person would use?
Gave it a small, friendly icon with three states: inactive, active, sniffing.
How can data visualization pull attention, stay readable, and still reward a closer look?
Glyphs: three small dials in the island, and a layered report one tap away.
How should risk look without making every post feel alarming?
Early explorations used a red, orange and green scale. The final uses red only, filled to show how much. Everything else stays black and grey, so a calm feed stays calm.