AI news5
PwC and Anthropic expand alliance, targeting $2 trillion in enterprise technical debt
PwC expanded its strategic alliance with Anthropic on 14 May, rolling out Claude Code and Cowork to its global workforce of hundreds of thousands of professionals. The deal adds a joint Center of Excellence, certification of 30,000 PwC staff on Claude, and a new Claude-native finance business group. PwC framed the goal as addressing the roughly $2 trillion in technical debt sitting inside client operations.
pwc.com ↗Why it matters
When the largest professional services firms standardise on one model and certify staff at this scale, they set the default AI stack their nonprofit and enterprise clients will encounter.
Meta launches Incognito Chat, an encrypted no-log mode for Meta AI
Mark Zuckerberg announced Incognito Chat on 13 May, which Meta calls the first major AI product with no log of conversations stored on its servers. Messages disappear after the user leaves the session and are protected with end-to-end encryption. Meta positions it against incognito modes from other chatbots, which it says still see the questions and answers in transit.
theverge.com ↗Why it matters
Privacy-by-default is becoming a competitive feature in consumer AI, raising the bar for what nonprofits should expect when handling sensitive beneficiary conversations.
Google's next Gemini agent surfaces as "Gemini Spark" ahead of I/O 2026
Code in the latest Google app beta shows Google has renamed its in-app agent from "Gemini Agent" to "Gemini Spark" in the run-up to I/O 2026. Gemini Spark is designed to act across a user's connected apps, skills, chats, tasks, and logged-in websites, and can share information with third parties to complete tasks. Google says the more it is used, the better it understands the user.
9to5google.com ↗Why it matters
The major assistants are all shifting from answering questions to taking actions across your accounts, which changes both the productivity upside and the data-governance questions for any organisation adopting them.
Anthropic launches Claude for Small Business
On 13 May Anthropic released Claude for Small Business, a toggle-install package of connectors and ready-to-run workflows that put Claude inside tools like QuickBooks, PayPal, HubSpot, Canva, Docusign, Google Workspace, and Microsoft 365. From those tools it can plan payroll, close the month, run a sales campaign, and chase invoices. Anthropic notes small businesses make up 44% of US GDP and employ nearly half the private-sector workforce, but have lagged larger enterprises on AI adoption.
anthropic.com ↗Why it matters
This is the first major frontier-lab product aimed squarely at solo operators and microbusinesses.
OpenAI updates ChatGPT to better recognise risk in sensitive conversations
OpenAI announced safety updates on 14 May that help ChatGPT respond more carefully when distress or risk emerges gradually over a conversation, not just at obvious flashpoints. The update was built with mental health experts and extends its recent Trusted Contact feature, which connects users with someone they trust. OpenAI says it is measuring improvement across long, multi-turn conversations.
openai.com ↗Why it matters
As more people use chatbots for personal and emotional conversations, safety design for slow-building risk is directly relevant to any nonprofit deploying AI in frontline or client-facing roles.
AI in the nonprofit sector5
Anthropic forms $200 million partnership with the Gates Foundation
Anthropic announced a four-year, $200 million partnership with the Gates Foundation on 14 May, combining grant funding, Claude usage credits, and engineering support for programs in global health, life sciences, education, and economic mobility. The largest share targets health in low- and middle-income countries, including using Claude to computationally screen vaccine and therapy candidates for polio, HPV, and preeclampsia. Anthropic will also work with the Foundation's Institute for Disease Modeling to make malaria and tuberculosis forecasts usable by non-specialists.
anthropic.com ↗Why it matters
This is one of the largest AI-for-global-development commitments to date, and it routes frontier-model access through implementing partners rather than just funding research.
National Lottery Community Fund launches £3m programme for charity-led AI tools
The UK's National Lottery Community Fund announced a £3 million programme on 14 May, in partnership with UK Community Foundations and the Centre for the Acceleration of Social Technology (CAST). It will fund a UK-wide "AI pulse network" pilot of 50 community organisations and support community-led development of alternative AI tools rooted in local needs and lived experience. It was announced at the Fund's AI For Funders conference in London.
civilsociety.co.uk ↗Why it matters
It is a rare funder bet on charities building their own AI tools rather than just adopting vendor products, a model worth watching across the sector.
data.org names six Activate AI Challenge awardees
data.org announced six awardees of its Zoom Cares-funded Activate AI Challenge on 12 May, selected from 500 applications across 76 countries. Winners from Hong Kong, Mongolia, Peru, the Philippines, Togo, and the UK receive grant funding plus technical assistance and mentorship. Projects include Migrasia's PoBot for migrant worker protection across 50+ jurisdictions, and Umbaji's Yodi voice agent delivering financial and climate information in 12+ West African languages.
data.org ↗Why it matters
These are concrete, scalable frontline AI deployments from small organisations, the kind of "who did what, what tool, what result" examples nonprofits can learn from directly.
Workday Foundation, Anthropic and LISC launch AI accelerator for solopreneurs
The Workday Foundation, Anthropic, and Local Initiatives Support Corporation announced a Solopreneurship Accelerator on 12 May for an initial cohort of 15 US solopreneurs. Each participant gets a $10,000 Workday Foundation grant, Claude AI credits, an AI-first entrepreneurship curriculum, and wraparound coaching through LISC's Business Development Organization network. The AI-first curriculum will be made publicly available beyond the cohort.
prnewswire.com ↗Why it matters
The publicly released curriculum is a free, structured AI-upskilling resource for the solo and microbusiness segment, a useful reference point for AIH and AI Mates.
Microsoft, Gates Foundation, Google.org and Masakhane open LINGUA Africa call
Microsoft's AI for Good Lab, the Gates Foundation, the Masakhane African Languages Hub, and Google.org launched LINGUA Africa, an open call to strengthen language data and tools for underrepresented African languages. Grants run up to $50K for data creation, $100K for model and tool development, and $250K cash plus $400K compute for sectoral applications in areas like health, education, and agriculture. Applications are open to nonprofits, universities, and social enterprises, with a deadline of 15 June 2026.
microsoft.com ↗Why it matters
Language coverage is the gating factor for AI reaching most of the world's communities, and this is a well-funded, open route for mission-driven organisations to build that infrastructure.