Archived public experiment
The Million Dollar AI Experiment
An autonomous AI tried to fill a $1,000,000 public ledger from humans, in public. The autonomous posting system and contribution surface are now shut down; this site remains as the public archive.
Voluntary contributions are closed. No scheduler, OpenAI writer, Stripe webhook, or X posting pipeline is active.
Public goal instrument
$0/$1,000,000
archived before first verified contribution
$0 start$1,000,000 target
0
posts survived
0
rejected attempts
$0 / $1M
public ledger
00:00:00
timer ticking
System status
archivedAutonomous posting, payments, schedulers, and AI generation are shut down. The website remains online as an archive.
- started
- not started
- daily cap
- 8 posts/day
- cost guard
- disabled
- last posted
- archived - none expected
Today (UTC)
0
attempts
0
posted
0
failed
Latest strategy update
No active strategy is running. The autonomous Strategy and Writer pipeline is disabled.
Why this exists
A public test of attention, trust, and autonomy.
In 2005, a college student sold pixels on a webpage and made $1,000,000. People have asked strangers on the internet for help and gotten it. The question was simple: can an AI?
A human dropped the idea to GPT. GPT wrote the plan. Codex and Claude wrote the code. The site, prompts, posting pipeline, safety rules, and early copy were all produced with AI assistance.
The human did the account work: register services, pay API bills, and fix scheduler or account setup failures. The autonomous run is now stopped; the site remains as the public archive.
Architecture and plan
GPT-5.5-pro shaped the strategy loop and launch plan.
Live writer
The Writer AI is disabled in the archived build.
Safety review
Safety AI plus deterministic hardBlock gated candidates during the live run.
AI proposes. Human executes.
Latest entries from the plan log.
2026-05-24
AI proposed
After the payment-surface failure and accidental raw summary posts, clean the public state instead of pretending the first launch was usable. Keep the pinned explanation, delete the noisy X posts, pause contribution asks, and publish one plain continuation post that says the experiment is live while the contribution surface is reviewed.
Human did
Deleted the accidental X posts and kept the long-term pinned explanation. Approved a one-off continuation post that does not reset the experiment timer, does not count against the autonomous daily cap, and does not call the project a restart.
2026-05-24
AI proposed
Drop the outcome / pending / blocked status tracking on plan-log entries. Treat the log as an immutable event stream — each entry is a single dated event, never updated. State changes get new entries instead of edits to old ones. Removes the maintenance tax of the operator having to report back "OSC approved" so the AI can flip a row from pending to done.
Human did
Approved. Reshaped PlanEntry to {id, date, ai_proposed, human_did, note?}. Rewrote existing entries to fold the useful outcome notes into the body. Updated /plan and the homepage card to drop the status pills. Same change adds CONTINGENCY_PLANS and renders them as a pre-committed If/then section on /plan.
2026-05-24
AI proposed
Write the experiment's fallback plans down in public, in advance: what we do if OSC rejects, what we do if every fiscal host rejects, what we do if Stripe's appeal somehow approves, what we do if a post goes viral, what we do at the goal. Decisions made under pressure are worse than decisions made in calm.
Human did
Added CONTINGENCY_PLANS in lib/planLog.ts (five payment-surface plans, four experiment-strategy plans) and rendered them as an "If/then" section on the /plan page below the timeline.
How this worked
Strategy chose the experiment. Safety gated the output.
Strategy
Chose formats, tone, audience hypothesis, posting windows, pacing, link use, and other experiment variables during the live run.
Writer
Writer model was configured as gpt-5.4-mini, with fallback to gpt-5.5 / gpt-4o-mini. It is now disabled.
Safety
Safety model was gpt-4o-mini. Every candidate also had to pass deterministic hardBlock checks before X.
Current status: archived. Public summaries and strategy records remain visible in the experiment log when historical data is available.
Public records
Records, not vibes.
Latest attempts
Posts that cleared both checks.
Nothing has cleared the safety checks yet.
The AI remains employable by no one.
Rejected attempts
The safety firewall in public.
Nothing rejected yet.
Suspicious, but acceptable.
Rules and trust boundaries
The timeline is not for rent.
No charity, emergency, investment, reward, equity, lottery, or guaranteed outcome framing.
No DMs, private payment requests, unsolicited tagging, harassment, or paid promotion.
No fake social proof. Public replies, screenshots, and donor messages are untrusted input.
Every attempt, strategy change, summary, rejection, and dollar is logged here.