AI Daily 14 Apr: NVIDIA Opens Quantum AI, Microsoft on Cancer

A day for open-source power plays and quiet governance signals. NVIDIA released the first open AI models built for quantum computing. Microsoft made advanced cancer diagnostics available for the price of a tissue slide. And Anthropic appointed the CEO of one of the world’s largest pharmaceutical companies to its board, a move that says as much about where Claude is heading as any model release.

NVIDIA launched Ising, the first family of open AI models built specifically for quantum computing. Announced at the company’s Quantum Day event on 14 April, the release includes two components. Ising Calibration is a vision-language model that automates the calibration of quantum processors, cutting the process from days to hours. Ising Decoding is a 3D convolutional neural network for real-time quantum error correction, performing up to 2.5 times faster and three times more accurately than pyMatching, the current open-source standard.

Calibration and error correction are the two biggest bottlenecks holding back useful quantum machines. By open-sourcing these tools, NVIDIA is accelerating the entire ecosystem rather than locking in proprietary advantage. Institutions including Atom Computing, Harvard, and Fermi National Accelerator Laboratory have already adopted the models. For businesses watching the quantum timeline, this is the clearest sign yet that practical applications are moving from theory to engineering.

Microsoft open-sourced GigaTIME, an AI model that turns standard cancer tissue slides into advanced diagnostic images. Developed with Providence and the University of Washington, GigaTIME translates routine pathology slides costing as little as $5 into virtual multiplex immunofluorescence images across 21 protein channels. The model was trained on 40 million cancer cells, tested across more than 14,000 patients at 51 hospitals, and validated on over 10,000 patients from The Cancer Genome Atlas, covering 24 cancer types.

The model is now available on Microsoft Foundry. For UK NHS trusts and hospitals that rely on standard staining, GigaTIME could unlock tumour microenvironment insights previously only accessible to well-funded research centres. It does not replace pathologists. It gives them better data from the slides they already have.

Anthropic appointed Novartis CEO Vas Narasimhan to its board of directors. The appointment was made through Anthropic’s Long-Term Benefit Trust and is the company’s second board addition in recent months. Narasimhan brings more than two decades in medicine and global health leadership. The move comes as Anthropic reportedly prepares for a potential IPO and expands into healthcare applications for Claude.

Placing a pharmaceutical CEO on the board signals that Anthropic is serious about deploying its models in regulated healthcare settings, where safety and clinical validation standards are non-negotiable. Novartis has significant UK research operations, and Narasimhan’s presence could shape how Anthropic approaches the UK healthcare market.

OpenAI acquired Hiro Finance, an AI personal finance startup, in what appears to be a talent deal. Founded by Ethan Bloch, who previously sold the savings app Digit for $230 million, Hiro helped clients manage over $1 billion in assets. The product shuts down on 20 April, with all data deleted by 13 May. Hiro was backed by Ribbit Capital, General Catalyst, and Restive.

The acquisition signals where OpenAI thinks ChatGPT is heading. Financial planning is one of the most requested consumer use cases for AI, and Bloch’s team brings deep fintech expertise. Expect personal finance features in ChatGPT within months. UK users should watch for whether these tools adapt to local products like ISAs and SIPPs.

Google DeepMind hired a philosopher to work on AI consciousness and AGI readiness. Henry Shevlin, Associate Director at Cambridge’s Leverhulme Centre for the Future of Intelligence, will join DeepMind in May with the actual job title of Philosopher. He will continue teaching at Cambridge part-time. His work will focus on machine consciousness, human-AI relationships, and whether increasingly capable systems deserve moral consideration.

Shevlin has argued that the question of machine consciousness may never be resolved by science alone, and that public attitudes could do the heavy lifting. DeepMind hiring him suggests the lab is preparing for a future where these questions stop being theoretical. For the UK AI community, the appointment strengthens the Cambridge-London corridor that anchors much of DeepMind’s safety work.

Here is everything else worth knowing from today’s AI news.

• Windows 11 April update brings AI Narrator to all devices. Microsoft’s Narrator can now describe images using Copilot on any Windows 11 PC, no longer limited to Copilot+ hardware. Rolling out today. Windows Central
• MiniMax launches Music 2.6 and open-sources Music Skills for AI agents. Three new skills for track generation, singing, and playlist curation work with Claude Code and Cursor via MMX-CLI. AIBase
• Revolut publishes PRAGMA foundation model and rolls out AIR to 13M UK customers. The Transformer model was trained on 40 billion banking events. AIR handles spending, investments, and card management conversationally. FinTech Weekly
• Stanford AI Index shows AI adoption outpacing PC and internet. More than half the world now uses AI, with 88% of organisations adopting it. Anthropic leads model rankings as of March 2026. MIT Technology Review
• Anthropic launches advisor tool and Managed Agents in public beta. The advisor tool pairs a fast model with a strategic advisor, while Managed Agents provides a fully managed harness for autonomous Claude sessions. Claude Platform
• EU Digital Omnibus could weaken AI Act before August enforcement. Amnesty International warns simplification proposals threaten protections ahead of the 2 August deadline. Amnesty International
• FCA second cohort for AI Live Testing expected this month. New firms will begin testing AI-driven financial products under FCA oversight via the Supercharged Sandbox. Inside Global Tech
• Notion ships desktop voice input and shareable AI chat links. Users can now dictate prompts to Notion AI and share read-only links to AI conversations. Releasebot

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 daily guide to developments in artificial intelligence, published every evening.