AI For Good

Adoption At A
Tipping Point

AI Adoption 2025 Report

AI Adoption 2025 Report

AI Adoption 2025 Report

Volume 3

Findings from 3 Sided Cube's 2025 AI adoption survey on how mission-aligned businesses, nonprofits and NGOs are using AI, right now.

Findings from 3 Sided Cube's 2025 AI adoption survey on how mission-aligned businesses, nonprofits and NGOs are using AI, right now.

Findings from 3 Sided Cube's 2025 AI adoption survey on how mission-aligned businesses, nonprofits and NGOs are using AI, right now.

Plus a practical playbook to do it responsibly!

Plus a practical playbook to do it responsibly!

Plus a practical playbook to do it responsibly!

89%

of organisations taking part said they had integrated AI to some extent

89%

of organisations taking part said they had integrated AI to some extent

89%

of organisations taking part said they had integrated AI to some extent

With 89% of purpose‑led organisations now on their AI journey, the question is no longer if - it’s how well we scale, govern and measure impact.
With 89% of purpose‑led organisations now on their AI journey, the question is no longer if - it’s how well we scale, govern and measure impact.
With 89% of purpose‑led organisations now on their AI journey, the question is no longer if - it’s how well we scale, govern and measure impact.

Tl:DR

Tl:DR

Tl:DR

Show me the data. A whistle-stop tour of what we found and what to do next.

Show me the data. A whistle-stop tour of what we found and what to do next.

Adoption accelerated

Non-adopters dropped to 11%; those fully integrated rose to 17%.

Adoption accelerated

Non-adopters dropped to 11%; those fully integrated rose to 17%.

Adoption accelerated

Non-adopters dropped to 11%; those fully integrated rose to 17%.

Real value is showing up

77% report noticeable improvements; 57% say they now see more successes than challenges.

Real value is showing up

77% report noticeable improvements; 57% say they now see more successes than challenges.

Real value is showing up

77% report noticeable improvements; 57% say they now see more successes than challenges.

Impact is trending positive

Positive project outcomes climbed to 38%; negative outcomes are falling.

Impact is trending positive

Positive project outcomes climbed to 38%; negative outcomes are falling.

Impact is trending positive

Positive project outcomes climbed to 38%; negative outcomes are falling.

Guardrails are catching up

56% now have an AI policy, but community engagement remains low.

Guardrails are catching up

56% now have an AI policy, but community engagement remains low.

Guardrails are catching up

56% now have an AI policy, but community engagement remains low.

Momentum continues

90% plan to deepen AI adoption in 12 months; overall confidence remains high (if a touch more realistic).

Momentum continues

90% plan to deepen AI adoption in 12 months; overall confidence remains high (if a touch more realistic).

Momentum continues

90% plan to deepen AI adoption in 12 months; overall confidence remains high (if a touch more realistic).

THE TIPPING POINT IS HERE

THOUGHTS FROM...

THOUGHTS FROM...

THOUGHTS FROM...

Duncan Cook
Duncan Cook

Founder & CEO

Founder & CEO

The tipping point is here. Now make it count.
This is our third annual pulse check on AI use in the purpose-led world. Adoption is now the default.

Most teams are using AI in day-to-day work and more report real improvements. The question is not if, it is how we make it useful, safe, and measurable.
At 3 Sided Cube, we are an AI-first product agency for good. We ship features that work in the wild, and we build responsibility into how we design, test and launch.

The pattern is simple. Start small. Prove it. Scale what works. Measure value in human terms. Keep people in the loop.
Thank you to everyone who shared their wins, worries and roadblocks. Use this report to pick one workflow, run a small test and share what changed. Then do it again.

Method

At A

Glance

WHat

Third annual AI adoption pulse - short survey on adoption, impact and governance.
1/4

WHat

Third annual AI adoption pulse - short survey on adoption, impact and governance.
1/4

WHEN

Fielded from July–August 2025.
2/4

WHEN

Fielded from July–August 2025.
2/4

Who

A diverse mix of decision‑makers and practitioners across charities, NGOs and mission‑aligned for‑profits globally.
3/4

Who

A diverse mix of decision‑makers and practitioners across charities, NGOs and mission‑aligned for‑profits globally.
3/4

How

Online survey; anonymised reporting.
4/4

How

Online survey; anonymised reporting.
4/4

Method

At A

Glance

WHat

Third annual AI adoption pulse - short survey on adoption, impact and governance.
1/4

WHEN

Fielded from July–August 2025.
2/4

Who

A diverse mix of decision‑makers and practitioners across charities, NGOs and mission‑aligned for‑profits globally.
3/4

How

Online survey; anonymised reporting.
4/4

Method At A

Glance

Adoption

Adoption

Adoption

The Tipping Point: where AI adoption stands now

The Tipping Point: where AI adoption stands now

The Tipping Point: where AI adoption stands now

'Non -adopters' in decline

% of respondents who said they weren’t using AI in some way

Depth is growing

17%

Full integration has increased 4.5x compared to 2024

Most teams are in progress

73%

are starting to integrate AI in day-to-day work

What does it
mean though?

The sector has crossed a tipping point. Adoption is the norm; the hard work shifts to quality of use, governance, and measurable value.

What does it
mean though?

The sector has crossed a tipping point. Adoption is the norm; the hard work shifts to quality of use, governance, and measurable value.

What does it
mean though?

The sector has crossed a tipping point. Adoption is the norm; the hard work shifts to quality of use, governance, and measurable value.

Value

Value

Value

Teams report concrete improvements, led by efficiency and productivity

Teams report concrete improvements, led by efficiency and productivity

Teams report concrete improvements, led by efficiency and productivity

Top 5 areas of improvement

What is one thing that has made an impact with AI this year?

Dafne Kiyui

ABillion - helping people eat healthier

0m 36s

Dafne Kiyui

ABillion

AI increasingly impacts project outcomes

38%

reported positive impacts

Up 11% from 2024

The numbers of those who report negative impact is declining year-on-year.

13%

13%

In 2025

24%

24%

In 2024

28%

28%

In 2023

Reality check

The story isn’t “AI fixed everything”. It’s more use, more value, and fewer misfires.

Reality check

The story isn’t “AI fixed everything”. It’s more use, more value, and fewer misfires.

Reality check

The story isn’t “AI fixed everything”. It’s more use, more value, and fewer misfires.

ETHICS &

GOVERNANCE

ETHICS &

GOVERNANCE

ETHICS &

GOVERNANCE

Policies vs reality

Policies vs reality

Policies vs reality

Ethics matter at the decision table

67%

say ethics concerns significantly shaped adoption

Why no policy (yet)?

Common reasons
✏️
It's in development…
It's in development…
It's too early…
It's too early…
🤷
It's not needed…
It's not needed…
🤹
We have competing priorities…
We have competing priorities…

Why no policy (yet)?

Common reasons
✏️
It's in development…
It's too early…
🤷
It's not needed…
🤹
We have competing priorities…

If you did one thing with AI in 30 days, what would it be?

Addie Achan

Impacts AI

0m 31s

Addie Achan

Impacts AI

Guardrails are catching up…

Guardrails are catching up…

What good
looks like

Good means having a clear purpose, strong human oversight, minimal and consent-based data use, and thorough vendor checks that consider geopolitical and data rights factors.

What good
looks like

Good means having a clear purpose, strong human oversight, minimal and consent-based data use, and thorough vendor checks that consider geopolitical and data rights factors.

What good
looks like

Good means having a clear purpose, strong human oversight, minimal and consent-based data use, and thorough vendor checks that consider geopolitical and data rights factors.

COMMUNITY

VOICE

COMMUNITY

VOICE

COMMUNITY

VOICE

The engagement gap: Community voice in AI work

The engagement gap: Community voice in AI work

The engagement gap: Community voice in AI work

Despite mission-driven values, community engagement is low

46%

rarely involve affected communities in AI projects

14%

engage with them often or very often

Where is one place you chose not to use AI, and why?

Jim Snape

National Deaf Children's Society

1m 16s

Jim Snape

National Deaf Children's Society

Why community engagement matters…

Why community engagement matters…

Reduces harm

Users catch risks you miss.

Reduces harm

Users catch risks you miss.

Reduces harm

Users catch risks you miss.

Build trust

Transparent, two-way design.

Build trust

Transparent, two-way design.

Build trust

Transparent, two-way design.

Improves adoption

Co-design results in real uptake.

Improves adoption

Co-design results in real uptake.

Improves adoption

Co-design results in real uptake.

Quick wins

Add community check-ins during key stages, budget for fair compensation, and share clear model cards and FAQs to support user understanding.

Quick wins

Add community check-ins during key stages, budget for fair compensation, and share clear model cards and FAQs to support user understanding.

Quick wins

Add community check-ins during key stages, budget for fair compensation, and share clear model cards and FAQs to support user understanding.

POTENTIAL

POTENTIAL

POTENTIAL

Confidence and the next 12 months

Confidence and the next 12 months

Confidence and the next 12 months

Confidence holds high

38%

express confidence in AI's potential to do good

Commitment rises

90%

plan to deepen integration in the next 12 months. 'Very Likely' is the fastest growing sentiment.

One key action

If you only do one thing, measure adoption in human terms. Track minutes saved, errors reduced, or people reached for one workflow. Report it monthly. No metric, no momentum.

One key action

If you only do one thing, measure adoption in human terms. Track minutes saved, errors reduced, or people reached for one workflow. Report it monthly. No metric, no momentum.

One key action

If you only do one thing, measure adoption in human terms. Track minutes saved, errors reduced, or people reached for one workflow. Report it monthly. No metric, no momentum.

READINESS

CHECKLIST

READINESS

CHECKLIST

READINESS

CHECKLIST

Before you dive into your AI project, run through this quick self-check to make sure the essentials are in place.

Before you dive into your AI project, run through this quick self-check to make sure the essentials are in place.

Before you dive into your AI project, run through this quick self-check to make sure the essentials are in place.

01

Define the problem

We’ve defined one problem worth solving and the user it helps.

01

Define the problem

We’ve defined one problem worth solving and the user it helps.

01

Define the problem

We’ve defined one problem worth solving and the user it helps.

02

Define KPIs

We know what “good” looks like (KPIs) before we start.

02

Define KPIs

We know what “good” looks like (KPIs) before we start.

02

Define KPIs

We know what “good” looks like (KPIs) before we start.

03

Ethical framework

We have a lightweight ethics/policy framework in place.

03

Ethical framework

We have a lightweight ethics/policy framework in place.

03

Ethical framework

We have a lightweight ethics/policy framework in place.

04

Community feedback

We’ve planned a community feedback step.

04

Community feedback

We’ve planned a community feedback step.

04

Community feedback

We’ve planned a community feedback step.

05

Define MVP

We can ship a pilot in < 8 weeks.

05

Define MVP

We can ship a pilot in < 8 weeks.

05

Define MVP

We can ship a pilot in < 8 weeks.

06

Escape hatch

We have a rollback plan if it under‑delivers.

06

Escape hatch

We have a rollback plan if it under‑delivers.

06

Escape hatch

We have a rollback plan if it under‑delivers.

BEYOND THE TIPPING POINT

Adoption is now

Adoption is now

the default

the default

With 89% somewhere on the curve and 17% fully integrated, the question isn’t if you use AI - it’s whether it’s making a measurable difference for people.
This year’s data says the promise is real (77% seeing improvements), but two gaps decide whether that promise sticks: guardrails and genuine engagement.

Three things that

Three things that

matter now

matter now

1
1
1
Prove value in human terms

Minutes saved, errors reduced, reach increased - pick metrics a programme lead would defend.

Prove value in human terms

Minutes saved, errors reduced, reach increased - pick metrics a programme lead would defend.

Prove value in human terms

Minutes saved, errors reduced, reach increased - pick metrics a programme lead would defend.

2
2
2
Tighten the guardrails

Policies exist for over half of orgs; make them practical: role-based access, audit trails, red‑teaming, rollback plans.

Tighten the guardrails

Policies exist for over half of orgs; make them practical: role-based access, audit trails, red‑teaming, rollback plans.

Tighten the guardrails

Policies exist for over half of orgs; make them practical: role-based access, audit trails, red‑teaming, rollback plans.

3
3
3
Bring communities in

With only 14% engaging often, co‑design is the fastest lever to reduce harm and increase uptake.

Bring communities in

With only 14% engaging often, co‑design is the fastest lever to reduce harm and increase uptake.

Bring communities in

With only 14% engaging often, co‑design is the fastest lever to reduce harm and increase uptake.