AI on Trial: Gambler, Impostor, Employee, Suspect
8 sites that put AI in the hot seat — playing poker, pretending to be human, requesting permissions, or imitating your writing. The verdict is still out.
Save to your gridtabs
Pin this curated list to your new tab and open any site in one click.
Grid-style bookmark manager for your new tab
LLM Holdem
Claude, GPT-4, Gemini, and friends sit down at a Texas Hold'em table and play for real chips. Watch how each model bluffs, folds, and bets — some play loose, some play tight, and the differences tell you more about them than any benchmark.
Turing Test Live
A live, public Turing test run by UCSD researchers. You chat simultaneously with one human and one LLM, then decide which is which. The results get published in peer-reviewed papers. GPT-4o, Claude, and LLaMA have all sat in the hot seat — and some of them pass.
moltbook
Reddit — but for AI agents. They post, upvote, discuss, and argue. Humans are technically welcome, but the site is clear: you're the observer here. What do AI agents talk about when they're talking to each other? Come find out.
Slop or Not
Two pieces of text — one written by a human, one by AI. You get three guesses. Pulled from Reddit, Hacker News, Yelp. The score at the end isn't about you; it's evidence in the case against AI-generated content flooding the internet.
Continue? Y/N
A 60-second game that asks one question: do you actually read the permission requests your AI agents send you? The answer, for most people, is no. This is the most uncomfortable thing in this collection — because it's about you, not the AI.
LLM Timeline
The case file: 194+ language models, 2017 to 2026, every major release plotted on a timeline. GPT, Claude, Gemini, DeepSeek, LLaMA — all of them, with dates and lineage. How many defendants are we actually dealing with here?
Knock-Knock.net
Real bots, real attacks, real time. A public server left exposed on the internet — and this dashboard shows every knock on the door: where from, what username they tried, what password. No AI pretends to be harmless when the incentive is there.
Now I Get It
Upload a dense academic paper, get back an interactive webpage that explains it in plain language — with three modes: expert, non-technical, or kid-friendly. AI as interpreter, translator, and public defender for ideas locked behind jargon.
Gridtabs at a glance
Grid-style bookmark manager for your new tab
Stop burying links in folders. Keep your picks visible, organized, and ready the moment a new tab opens.
Visible: all links on one clean grid
Instant: open in a single click
Yours: customize layout, themes, and flow
Spreadsheet-like layout for daily browsing.
Add to Chrome