Saturday roundup. Open models getting seriously capable, Microsoft going all-in on nonprofits, and a growing conversation about whether AI’s environmental cost is something the sector needs to reckon with.
Google Gemma 4: open-weight models go fully multimodal
Google released Gemma 4 under Apache 2.0. Models from 2B to 31B parameters with 256K context, native vision and audio, 140+ languages, and built-in function calling. Runs on consumer hardware.
Nonprofit angle: Free, open models that handle text, images, and audio in 140+ languages. If your org works across multiple languages or needs to process visual content on a budget, this is worth looking at seriously.
Source: Asanify
Microsoft Nonprofit Leadership Summit: $5B+ commitment
1,500 organisations gathered in Bellevue for Microsoft’s Global Nonprofit Leadership Summit. The centrepiece: a $5B+ commitment to nonprofit tech support, focused on AI adoption and agentic tools.
Nonprofit angle: If your org uses Microsoft tools (most do), this matters. The investment is specifically targeting AI adoption support for nonprofits. Worth tracking what programs come out of this.
Source: GeekWire
AI’s environmental cost is getting harder to ignore
Cornell researchers project US data centres will produce CO2 equivalent to 5-10 million cars and consume water equivalent to 10 million Americans by 2030. Growing concern among nonprofit leaders about the sustainability trade-off.
Nonprofit angle: If your organisation works on environmental issues (or just cares about its footprint), this is a conversation worth having internally. Using AI responsibly includes understanding its environmental cost.
Source: GeekWire
IBM + NationSwell: Responsible AI for Social Impact Playbook
A practical guide for nonprofits, funders, and universities on ethical AI adoption. Covers safety, usefulness, and real-world implementation. Not theory, actual guidance you can use.
Nonprofit angle: This is genuinely one of the better resources for nonprofits figuring out AI governance. Worth downloading and sharing with your board or leadership team.
Source: IBM Newsroom
Anti-AI marketing campaigns are emerging
Brands like Aerie, Almond Breeze, and Equinox are actively rejecting AI-generated content. Two-thirds of consumers now question content authenticity. Human-created content is becoming a competitive differentiator.
Nonprofit angle: Your stories are real. Your impact is real. That authenticity is genuinely becoming more valuable, not less. If you have been worried about competing with AI-polished content from bigger orgs, this trend suggests the opposite is happening.
Source: Neural Buddies
The IBM playbook is a good starting point if your org has not nailed down its AI governance yet. What’s your biggest question about adopting AI responsibly?