Support The AI Risk Network

$413 billion to build it. About $7,000 a month to explain it.

Meta, Alphabet, Amazon and Microsoft spent about $413 billion on capital expenditure in 2025, much of it on AI infrastructure. The AI Risk Network, the weekly shows that explain what that means for everyone else, runs on roughly $6,000 to $8,000 a month, and it depends on donations to keep going.

Source: Statista, July 2026. Running costs are Guard Rail Now's own figures.

Latest episode

Who’s in Control? AI Safety, Self-Improvement, and Surveillance

80k-150k

views a week across YouTube, TikTok, Instagram, X and Substack

weekly shows: Warning Shots and For Humanity

~$15

funds about 1,000 views of AI safety content

$6k-8k

total monthly cost to run the network

About $230 keeps The AI Risk Network running for one day. About $1,600 covers a full week.

How we got these numbers: midpoint cost of $7,000 a month, about 500,000 views a month (115,000 a week on average). That works out to roughly $14 per 1,000 views; we round up to $15 so we never overstate what a gift does.

Why this cause

A serious risk, very little funding, and a public that can still change the outcome.

Three reasons we think explaining AI risk to the public is one of the most useful places a small gift can go.

Big

38% to 51%

of 2,778 published AI researchers gave at least a 10% chance that advanced AI leads to outcomes as bad as human extinction, in the largest survey of its kind. In 2023, Geoffrey Hinton, Yoshua Bengio and the heads of the leading AI labs signed a statement calling the risk a global priority alongside pandemics and nuclear war.

Sources: Grace et al., Thousands of AI Authors on the Future of AI (2024); Center for AI Safety, May 2023.

Neglected

~20x less

philanthropic money went to AI safety and security in 2024 than to climate, according to estimates from Coefficient Giving, one of the largest funders in the field. Meanwhile the four largest tech companies are expected to spend about $760 billion on capital expenditure in 2026.

Sources: Coefficient Giving; Statista, July 2026.

Solvable

99 to 1

was the US Senate vote in July 2025 to remove a proposed 10-year ban on state AI laws, after public pushback. That September, California signed SB 53, the first US state law requiring developers of the most powerful AI models to publish safety frameworks and report serious incidents. Public attention moves policy.

Sources: States Newsroom, July 2025; Office of the Governor of California, Sept 2025.

Where your gift goes

Production, editing and distribution of the shows people actually watch.

Your gift pays for recording, editing, clips and publishing across every platform we're on, so the conversation about AI risk reaches people outside the tech world.

Weekly roundtable

Warning Shots

John Sherman, Liron Shapira of Doom Debates and Michael of Lethal Intelligence go through the week's AI news and what it means for AI risk, in plain language.

Watch on YouTube

Weekly interview

For Humanity

John Sherman sits down with researchers, organizers, parents and policymakers to talk about AI, its risks, and what people can do about them.

Watch on YouTube

Short form

Clips and shorts

The most important minutes of each episode, cut for TikTok, Instagram, X and YouTube Shorts. This is where most new viewers first find us.

See the clips

Writing

Substack and the blog

Sourced write-ups of every episode, so people can check the facts behind what they heard and share them with others.

Read on Substack

What your gift does

See your monthly gift in views and days on air.

Your monthly gift

$25

Estimates use $15 per 1,000 views and $230 per day of running costs. Views are counted across all platforms, and one person can watch more than once.

1,700views of AI safety content funded every month

20,000views funded over a year

1.3days of the whole network kept on air each year

If 280 people gave $25 a month, the entire network would be covered.

From the shows

In their own words

"Stop building what we don't understand."

John Sherman

Host, Warning Shots and For Humanity

"We do not have to agree about why to agree about the solution."

Michael Trazzi

Organizer, on For Humanity

"More activity from lawmakers in 24 hours than I've seen since the advent of AI."

Jim VandeHei

Co-founder of Axios, on For Humanity

Two ways to support

Give directly, or become a paid subscriber.

Both keep the network running. Pick the one that suits you.

Donate through Every.org

A direct gift

Tax-deductible in the US (501(c)(3))
One-time or monthly, any amount
Receipt sent by email right away

Choose an amount

Subscribe on Substack

Paid subscription

In-depth, fully sourced posts and articles only paid subscribers can read
Every episode written up, with the sources behind it
Monthly or annual plans
Subscriptions aren't tax-deductible

Become a paid subscriber

Questions

Before you give

Is my donation tax-deductible?

Yes, in the US. Guard Rail Now is the public name of Dads Against AGI, a 501(c)(3) nonprofit (EIN 99-4857285). Every.org sends you a tax receipt by email right after you give.

Why do you use Every.org?

Every.org is a nonprofit donation platform. It handles secure payment and receipts, so all of our time goes into the shows instead of running a payment system.

Can I change or cancel a monthly gift?

Yes, anytime. Monthly gifts are managed from your Every.org account, where you can change the amount or cancel.

How do you calculate the impact numbers?

The network costs $6,000 to $8,000 a month and gets 80,000 to 150,000 views a week across all platforms. We use the midpoints, which gives about $14 per 1,000 views, and round up to $15. These are estimates, not guarantees, and we update them as our costs and audience change.

What's the difference between donating and a paid Substack subscription?

Both fund the same work. A donation through Every.org is tax-deductible in the US. A paid Substack subscription isn't, but it gives you access to posts and articles that only paid subscribers can read.

Can I help without giving money?

Yes. Subscribe to The AI Risk Network, share an episode with one person who hasn't thought about AI risk, and take action through the resources on our site.

When I started making these shows, I didn't know if anyone outside the AI world would care. Now 80,000 to 150,000 people a week watch, share and argue with what we put out, and many of them had never thought about AI risk before they found us.

We don't have a lab's budget, and we don't need one. We need enough people giving a little every month to keep the cameras on. If you've learned something from us, I'd be grateful if you'd help keep it going.