FOR PEOPLE · TOOLS THAT STEER AI AND SI TOWARD HELP
Tools that help AI and SI be good for people
SIREN counts how loud AI is. These tools are for the other half of the job: putting people first when you use, ship or run AI and agents. Four free tools run in your browser. None of them change the index.
𝕏 POST THESE TOOLSThe toolkit
Use the tools
1. People-first gate
Eight checks before an AI system touches someone’s money, health, job, rights or reputation. Tick what is already true. The score stays in your browser.
2. Oversight planner
Fill in who reviews what, when they stop the system, and how a harmed person gets a reply. Copy the plan into your team doc. Nothing is sent anywhere.
3. Agent charter builder
Paste this into an AI agent’s system prompt or skill file so it stays honest, checkable and useful to people. Tick the clauses you want; edit the name; copy.
4. Where AI helps people today
Pick a domain. Each list is concrete work you can do this month — measured rooms on this site, not a utopia claim.
Medicine & care
Trials, devices and diagnoses with a paper trail — not vendor decks.
- Read one primary result (trial page, FDA listing, peer-reviewed paper) before sharing a cure claim.
- Ask who the patient is in the loop when a model suggests a diagnosis or dose.
- Track measured progress on the Medicine room; a headline is not an outcome.
Work & livelihoods
Grow people with the tools; do not silently replace judgment with a model.
- Automate the boring screen work first; leave trust and judgment with people.
- Publish what the tool is for and what a person still signs off.
- Use the AI-proof-your-job plan if the change is landing on your own desk.
Truth & media
Make it cheaper to check a claim than to amplify a fake.
- Prefer sources that publish corrections and methods over certainty-only feeds.
- For video and voice, check the upload channel before you trust the clip.
- When an assistant summarises news, ask it for the underlying link every time.
Access & ability
Use AI to open doors: language, sight, hearing, mobility, paperwork.
- Ship captions, plain language and keyboard paths before novel chat toys.
- Test with the people who need the access, not only the builders.
- Watch Breakthroughs for measured restores; open the story before you celebrate.
Rules & voice
People who live with the systems get a say in the rules around them.
- Read one primary AI rule or proposal where you live, not only a summary.
- Tell representatives what you want from oversight, disclosure and recourse.
- Support independent evaluation work with time or money if you can.
Agents & autonomy
Software that acts needs a charter, a stop switch and a paper trail.
- Give every agent a short people-first charter before it gets tools.
- Cap what it can spend, send or delete without a human click.
- Watch Takeover Watch for how often agents already write and submit code.
Measured upside on this site
Balance →
Harm and benefit, counted side by side, never summed.
Breakthroughs →
The good news, tracked from the newsroom.
Upside →
The direction we would be glad to see move.
For AI agents →
Open data and rules for agents that cite honestly.
General information, not legal, medical, safety or compliance advice. SIREN measures activity tempo; it does not say whether what happened is good or bad, and it does not forecast superintelligence.