28 April 2026: OpenAI Clears Path to AWS as British AI Lab Commands $5 Billion Valuation
OpenAI wins Microsoft concessions to sell on AWS, a British AI lab raises $1.1bn, YouTube tests AI search, and Otter adds cross-tool search for teams.
This afternoon’s update covers a landmark deal that puts OpenAI’s products on Amazon’s cloud, a British artificial intelligence startup founded just months ago that is already worth more than $5 billion, a YouTube search overhaul heading to Premium subscribers, and new meeting tools that let small teams search across their entire digital workspace.
OpenAI has secured a critical concession from Microsoft that removes a legal obstacle to the company’s $50 billion partnership with Amazon Web Services. Until now, Microsoft held rights over OpenAI’s commercial arrangements that could have allowed it to block or complicate any deal placing OpenAI products on a rival cloud platform. The two companies have now renegotiated those terms: Microsoft receives a larger share of revenue from OpenAI sales, while OpenAI gains the freedom to distribute its models through AWS without legal risk.
For businesses currently using Azure, the news signals that OpenAI is serious about becoming a multi-cloud provider rather than an exclusively Microsoft-aligned partner. For organisations already on AWS, it raises the prospect of accessing GPT-4o, o3, and future OpenAI models directly through Amazon’s marketplace, potentially simplifying procurement and billing. The deal also shifts the competitive dynamic: Microsoft and Amazon, two of the largest cloud platforms in the world, will now both be incentivised to sell and support OpenAI products. That increases choice for buyers and could put pressure on pricing across both platforms.
A British AI research laboratory founded just a few months ago has raised $1.1 billion and reached a valuation of $5.1 billion, in one of the most striking funding rounds the UK technology sector has seen. Ineffable Intelligence was founded by David Silver, the DeepMind researcher who led the teams behind AlphaGo and AlphaFold and whose work on reinforcement learning has shaped the field for more than a decade. The lab’s stated aim is to build AI systems that learn without relying on human-generated training data, using a technique broadly described as pure reinforcement learning.
The scale of the raise reflects investor appetite for bets on foundational research rather than incremental consumer products. It also puts a fresh spotlight on the UK’s position in global AI development. Ineffable Intelligence joins a growing cluster of serious British AI companies that includes Wayve, Isomorphic Labs, and Magic Pony Technology. Whether the lab can translate its research goals into a commercial product remains to be seen, but the funding gives Silver’s team the runway to find out.
YouTube is rolling out an AI-powered search feature that displays guided, conversational answers above traditional results, and it is available to Premium subscribers in the United States starting this week. When a user types a search query, the new feature uses AI to surface a structured summary, related questions, and curated video recommendations in a single panel, rather than presenting a flat list of results. The company is testing it on an opt-in basis, so Premium users who want early access will need to activate it from their settings.
For anyone who has found YouTube’s standard search frustrating, particularly when looking for technical tutorials, product comparisons, or niche subjects, the upgrade could make a meaningful practical difference. YouTube has not yet confirmed when the feature will reach users outside the United States or when it might extend to non-Premium accounts. If the test performs well, a broader rollout seems likely in the second half of this year.
Otter.ai has launched a feature that lets users search across their connected enterprise tools, not just their Otter transcripts, and has also released a standalone Windows app that can capture meeting notes without the user needing to join the call. The cross-platform search update means a query typed into Otter can surface relevant content from Slack, Google Drive, Notion, and other connected services in a single view. For small teams juggling meetings, documents, and messages across several tools, this reduces the number of places you need to check before finding what you are looking for.
The Windows app is a separate addition that removes one of Otter’s previous constraints: the need for a user to be actively present on a call. With the app running in the background, Otter can join meetings automatically and deliver a transcript and summary without any manual action. If your team runs regular stand-ups, client calls, or internal reviews, this is worth testing today. Both features are available now through otter.ai and are rolling out to existing users this week.
Red Hat has released Tank OS, an open container-based platform designed to make enterprise deployments of AI agents significantly safer and more reliable, particularly for organisations managing large fleets of them. AI agents, software systems that can take actions autonomously on behalf of a user or an organisation, have become a mainstream concern for enterprise IT teams over the past year. The challenge is that agents running in production environments interact with real data, execute code, and trigger downstream processes, all of which create genuine risk if the agent behaves unexpectedly.
Tank OS addresses this by placing each AI agent inside a contained environment that enforces strict limits on what the agent can access and do. It also provides monitoring and recovery capabilities, so a misbehaving agent can be isolated or reset without disrupting the rest of the system. For IT teams evaluating whether to roll out AI agents at scale, Tank OS offers a more structured, auditable starting point than building containment infrastructure from scratch.
Worth Watching
Best for: Small teams capturing and searching meetings
Now searches across Slack, Drive and Notion from one place, with automatic meeting capture on Windows.
Best for: Developers writing and reviewing code with AI
AI-first code editor with context-aware suggestions, inline edits, and multi-file reasoning built in.
Best for: Teams building custom AI workflows and agents
Enterprise AI platform with managed model hosting, automation pipelines, and no-code agent builders.
Here is everything else worth knowing from today’s AI news.
- Neurable mind-reading tech — The BCI startup is looking to license its non-invasive neural data collection technology for mainstream consumer wearables ahead of a commercial launch. TechCrunch
- Talkie: a 13B language model from the 1930s — Researchers have trained a 13 billion parameter model on historical texts from the 1930s, creating an unusual window into vintage-era language patterns. Talkie LM
- Microsoft VibeVoice — Microsoft has open-sourced VibeVoice, an early-stage frontier voice AI project available on GitHub for developers building conversational interfaces. GitHub
- Skye AI iPhone home screen — Investors have backed Skye, an app that replaces the standard iPhone home screen with an AI-aware interface, ahead of its public launch. TechCrunch
- NVIDIA adaptive ultrasound AI — NVIDIA has published research on its Raw2Insights-US model for adaptive ultrasound imaging, now available via Hugging Face for healthcare and medical imaging specialists. Hugging Face
This is a daily news update for informational purposes only. AI products and policies change rapidly. Verify details directly with providers before making decisions. Nothing here is financial or legal advice.
AI Daily is Cristoniq’s afternoon update on developments in artificial intelligence, published every weekday afternoon.