What if the same companies building the AI in your child’s pocket are also the ones deciding whether it’s safe?
That was basically the deal for years. A few companies built the tools, wrote their own rulebook, and told the rest of us to trust them.
Last year the pushback got organized. Parents, researchers, investors, even former presidents and prime ministers started running real campaigns to put limits on AI.
Some go after the technology. Some are aimed squarely at protecting kids.
Here’s who’s doing the work, and what they actually want.
Drawing red lines before things go wrong
In September 2025, the journalist Maria Ressa stood up at the UN General Assembly and said plainly what a lot of people had been circling for months.
The world has to decide what AI should never be allowed to do. Her speech launched the Global Call for AI Red Lines.
The list of names behind it is long, and not the usual crowd of tech people. Over 200 public figures and 70 organizations have signed so far, including ten Nobel winners and nine former presidents and ministers.
A few you’d probably recognize: Yoshua Bengio and Geoffrey Hinton, two of the researchers who built the groundwork for modern AI, plus the historian Yuval Noah Harari and the actor Stephen Fry.
What they’re asking for is plain enough. They want governments to agree on real, enforceable limits for the worst uses of AI by the end of 2026.
And this isn’t about some far-off science fiction scenario. They point to damage that has already been done, including teenage lives lost in connection to chatbot use.
We’ve drawn lines like this before, around biological weapons and human cloning, and the case here is that AI now sits in the same company.
Three things AI should never do to children
If you read about one campaign here, read about this one.
In October 2025, the Safe AI for Children Alliance launched its Global AI Non-Negotiables Campaign. Instead of wading into the bigger arguments about whether AI is good or bad, it names three things AI should never do to a child:
- AI must never create fake sexualized images of children.
- AI must never be designed to make children emotionally dependent.
- AI must never encourage children to harm themselves.
The campaign wants parents and teachers to write to their elected officials, and it hands you everything to do it: a template letter and a tool to find your local representative.
SAIFCA makes the case that a letter from an actual constituent moves a politician far more than one more signature on a petition. The whole thing takes about ten minutes.
A design code that puts children first
Baroness Beeban Kidron has spent years pushing tech companies to account for young users. In March 2025, her organization 5Rights released the Children & AI Design Code.
The Code is not a call to slow AI down. It asks developers to build systems with children’s safety and development in mind across the full life of the product, from the first design choice to deployment.
Kidron’s point is blunt: we keep leaving it to a few dominant companies to decide what counts as our children’s best interests, and that keeps going wrong.
The investors applying pressure with money
Campaigns work better when they hit the bottom line.
By late 2025, the Collective Impact Coalition for Ethical AI had 64 investors behind it, together managing USD 11.3 trillion, according to its 2025 progress report.
They use that weight to push the tech companies they invest in for clear AI policies and real governance, the kind you can actually check. It started with 44 members in 2022 and has grown to 78 today, 64 of them investors.
When the people holding the shares start asking hard questions, boardrooms listen.
Voting with your AI account
The Boycott Careless AI movement takes a consumer approach. It uses the Future of Life Institute’s AI Safety Index to rank the major labs, then asks people to use only the safety leader and avoid the rest until they catch up.
The Winter 2025 Index makes the wider problem clear. The Future of Life Institute graded every major lab, and the top score was a C+.
Its own summary states that no company is doing well. That tells you how far the whole field still has to go.
A few more worth a bookmark
A few others are worth knowing about:
- Amnesty International put out an Algorithmic Accountability toolkit, a set of tools for activists and reporters who want to investigate how AI systems harm people’s rights in practice.
- Humanists International agreed on the Luxembourg Declaration, a set of ten ethical principles for how AI should treat people.
- Sony AI released FHIBE, a public dataset for testing whether computer vision systems treat people fairly.
- Day of AI and MIT RAISE launched Responsible AI for America’s Youth, a movement to teach students and teachers how AI actually works.
What you can do this week
You don’t need to wait for a new law to act.
Check which AI tools your child uses, and whether they include child safety settings. Turn them on.
Talk to your kids about how these apps are built to keep them scrolling. They should know the pull they feel was designed by adults who get paid for every extra minute.
And if you want one principle to hold onto: a company that won’t tell you how its AI handles children is telling you something already.
Sources
Responsible AI for America’s Youth, Day of AI and MIT RAISE (September 25, 2025): prweb.com
Global Call for AI Red Lines, The Future Society (September 25, 2025): thefuturesociety.org
Global AI Non-Negotiables Campaign, SAIFCA (October 17, 2025): safeaiforchildren.org
Children & AI Design Code, 5Rights Foundation (March 18, 2025): 5rightsfoundation.com
Collective Impact Coalition for Ethical AI 2025 Progress Report, World Benchmarking Alliance (December 10, 2025): worldbenchmarkingalliance.org
Boycott Careless AI, built on the FLI AI Safety Index (Winter 2025): boycottcarelessai.org
Algorithmic Accountability toolkit, Amnesty International (December 9, 2025): amnesty.org
Luxembourg Declaration on AI and Human Values, Humanists International (July 10, 2025): humanists.uk
FHIBE Fairness Evaluation Dataset, Sony AI (November 5, 2025): sony.mediaroom.com