The Daily AI Briefing

The Daily AI Briefing — 28 April 2026

AI news5

  1. OpenAI ends Microsoft exclusivity: AWS and Google Cloud now in play, Microsoft revenue share preserved through 2030

    OpenAI and Microsoft announced an amended partnership on 27 April that ends Microsoft's exclusive license to OpenAI's models. Microsoft retains a non-exclusive IP license through 2032 and revenue share through 2030, but OpenAI can now sell on AWS and Google Cloud. Amazon CEO confirmed OpenAI models are coming to AWS soon. The restructure clears the legal path for OpenAI's $50 billion Amazon deal.

    Why it matters

    The single largest enterprise-AI distribution choke point of the past two years just opened up, which means every customer that bought into the "Azure-only" framing now has a vendor-portability conversation to have.

    blogs.microsoft.com
  2. David Silver's Ineffable Intelligence raises $1.1B seed at $5.1B valuation to build AI without human data

    Ex-DeepMind reinforcement learning head and AlphaGo lead architect David Silver announced a $1.1 billion seed round for Ineffable Intelligence on 27 April, valuing the months-old London lab at $5.1 billion. Backers include Nvidia and Google. Silver's stated thesis is that the era of training on human data is ending and that reinforcement-learning-only systems will overtake current frontier models.

    Why it matters

    This is the largest European AI seed ever, the first frontier-scale lab to publicly bet against human-data training, and a credibility re-anchor for London as a frontier-AI city.

    cnbc.com
  3. Apple iOS 27 will let third-party AI assistants run inside Siri, including Gemini, Claude, and ChatGPT

    Reports on 27 April confirmed Apple is building an extension framework into iOS 27 that will allow users to wire third-party AI assistants directly into Siri, including Google Gemini, Anthropic Claude, and OpenAI ChatGPT. iOS 27 previews at WWDC in June with September release. This extends Monday's confirmed Apple-Google $1B/yr Gemini deal into a multi-model architecture.

    Why it matters

    Apple is moving from "rent one frontier model" to "let users compose their assistant", which turns the iPhone into a neutral runway for every frontier lab and hands real power back to whichever assistant the user trusts.

    macrumors.com
  4. OpenAI smartphone with AI agents replacing apps: Qualcomm and MediaTek named on the custom processor, Luxshare on assembly

    Reports on 27 April confirmed OpenAI is building an AI-native smartphone with Qualcomm and MediaTek jointly designing the custom processor and Luxshare exclusively manufacturing the device. The architecture is reportedly designed around AI agents replacing apps as the primary interaction layer, building on the existing Jony Ive and Sam Altman hardware collaboration.

    Why it matters

    The supply-chain detail makes this the first concrete evidence that the post-app era is moving from speculation to silicon, and it sets up a direct platform fight with Apple inside the next 18 months.

    thenextweb.com
  5. Germany's Sereact raises $110M Series B for Cortex 2.0 "robotic brain" and US expansion

    Sereact announced a $110 million Series B on 26 April to scale Cortex 2.0, the company's foundation model for robotic manipulation, and expand into the United States. Cortex is positioned as a domain-general "brain" layer that lets industrial robots reason about novel objects and tasks without retraining. The round follows Sony AI's Project Ace robot table-tennis milestone last week.

    Why it matters

    A second well-funded foundation-robotics company outside the US, scaling on the heels of a competitive-sport robotics milestone, is a real signal that embodied agents are now an investable category not just a research bet.

    pymnts.com

AI in the nonprofit sector4

  1. Fast Forward kicks off Round 2 of the AI for Humanity Report with a fresh sector survey

    Fast Forward announced research has opened for the 2026 AI for Humanity Report, the second annual study of how nonprofits are building, funding, and deploying AI for social impact. The 2025 round surveyed nearly 200 organisations and found 30% of AI-powered nonprofits operate on budgets under $500K. This year's survey is open to nonprofits and funders worldwide.

    Why it matters

    The AI for Humanity Report is becoming the canonical sector reference for nonprofit AI adoption, and round-two participation is how nonprofits get their actual deployment story counted in the data funders use.

    linkedin.com
  2. Nature: agentic AI is about to break grant-funding systems, and funders need to act now

    Nature published a commentary on 27 April warning that agentic AI tools are about to flood grant-funding programmes with high-quality applications at a volume that breaks current human-review pipelines. The piece argues funders must adopt structural changes (explicit AI-disclosure rules, verification workflows, capped submission velocities) before the next major funding cycle.

    Why it matters

    Every nonprofit considering AI-assisted grant writing should read this now, because the funder-side response is going to reshape what "valid submission" means inside 12 months.

    nature.com
  3. Watershed launches sustainability AI agents and an 8-week AI Fellowship for sustainability leaders

    Watershed announced new sustainability-specific AI agents on 21 April that cut data-cleaning time by 80-93% across pilot customers, plus the Watershed AI Fellowship, an 8-week accelerator running 12 May to 30 June 2026 for sustainability professionals. Royal Mail and Smiths Group reported saving 12 weeks per year and collapsing weekly reporting cycles to a single day with the agents.

    Why it matters

    This is the first verticalised AI agent product with concrete time-saving numbers and a structured capacity-building pipeline behind it, which gives every nonprofit-vertical (case management, fundraising, programmes) a working template to demand from their tooling.

    watershed.com
  4. Google.org commits $10M to Manufacturing Institute to train 40,000 US manufacturing workers in AI

    Google.org announced a $10 million commitment to the Manufacturing Institute on 15 April to fund AI training programmes for 40,000 current and aspiring manufacturing workers across 15 US regions, with apprenticeship pathways included. The Manufacturing Institute is the workforce-development nonprofit arm of the National Association of Manufacturers.

    Why it matters

    Workforce-AI capacity-building is becoming the standard line item in big-tech philanthropy, and the Manufacturing Institute model (sector-trade-association partnership) gives other nonprofit-trade-associations a clear pitch to bring to funders.

    manufacturingdive.com

Free, and actually free

Courses, templates and working tools for nonprofits

No card, no trial. The Briefing publishes here every morning.

Open the Hub →