19 May 2026: Anthropic Acquires Stainless, Musk Loses OpenAI Trial (AM)

Anthropic has moved to own the company that builds the software connectors underpinning the entire AI industry, while a California jury ended Elon Musk’s two-year legal battle against OpenAI in under two hours. Elsewhere, AI is now making podcasts on demand, drug discovery models are opening up to non-scientists, and enterprises are getting new options to run coding agents on their own infrastructure.

Anthropic has acquired Stainless, the New York startup that built the official software development kits for OpenAI, Google, Cloudflare, and dozens of other technology companies. Stainless, founded in 2022, specialises in automating the creation and maintenance of SDKs: the code libraries developers use to connect their applications to an API. Without well-maintained SDKs, every company building on an AI platform must write and maintain that integration layer itself, which is slow and costly. Stainless solved that problem and became the quiet backbone of how AI products embed into real software.

The deal is reported to be worth at least $300 million, roughly double Stainless’s late-2024 valuation. For Anthropic, this means the team that already understands the Claude API better than anyone is now internal rather than external. The bigger question is whether Stainless will continue to maintain the SDKs it built for OpenAI and Google now that it belongs to their main rival. Anthropic has not yet said how that arrangement will work, and that answer is coming soon.

A California jury took less than two hours to reject all of Elon Musk’s claims against OpenAI CEO Sam Altman, finding that Musk had filed his lawsuit too late under the statute of limitations. After three weeks of testimony and hundreds of documents, the nine-person jury dismissed the case on procedural grounds, without ruling on whether Musk’s underlying claims about OpenAI’s non-profit-to-for-profit pivot had any merit.

Musk’s lawyer announced an immediate appeal. The practical impact for the AI industry is significant: OpenAI has cleared a major obstacle ahead of its planned IPO and holds its $40 billion in recent funding without the threat of a court-ordered restructuring. Whether the governance questions raised during the trial get answered in a higher court or in public debate is the next thing to watch.

SandboxAQ has brought its drug discovery models into Claude, making specialised scientific AI accessible without requiring a computing background. The company, which spun out of Google in 2022, has built models designed to identify promising drug compounds. Integrating them into Claude means researchers can query those models through a plain-language interface, without needing to write code or understand the underlying architecture. SandboxAQ is betting that the main obstacle to AI adoption in science is access, not capability. Other competitors like Chai Discovery and Isomorphic Labs have focused on building better models; SandboxAQ is focused on removing friction for the people who already know what they need. If you are curious how AI tools are being applied across professional settings, our piece on AI decision support covers the broader trend.

OpenAI and Dell have announced a partnership to deploy Codex, OpenAI’s AI coding agent, in hybrid and on-premise enterprise environments. Codex has so far run as a cloud-hosted product, meaning the code it processes travels through OpenAI’s infrastructure. For organisations with strict data residency or security requirements, this has been a barrier. The Dell partnership gives those organisations a route to run Codex on hardware they control, closing the gap between AI capability and corporate compliance requirements.

For smaller businesses and individual developers, this matters less immediately. But when the largest companies deploy AI coding agents as standard enterprise infrastructure, the tooling and expectations they establish tend to filter down to smaller teams within a year or two. Our guide to AI PCs and what they do differently is a useful primer if you are still getting to grips with how AI is changing hardware decisions.

Amazon’s Alexa+ can now generate custom podcast episodes on demand, turning its AI assistant into a personalised audio content platform. Users can request an episode on any topic, and Alexa+ will research, script, and produce a short audio piece. The feature is available now in the United States. UK availability has not been confirmed, which is a recurring frustration for British users of Amazon’s AI products. The direction is clear regardless: voice AI is moving from content retrieval toward content creation, and the distinction between using an AI tool and consuming AI-generated media is narrowing fast.

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

• EU AI Act high-risk rules delayed by 16 months: European legislators agreed a Digital Omnibus deal that pushes Annex III compliance obligations back to December 2027, from August 2026. High-risk systems in hiring, healthcare, and education now have significantly more time to comply. UK businesses operating in EU markets should update their compliance timelines accordingly. [18 May]
• Cursor Composer 2.5 released: Cursor shipped a new version of its Composer coding assistant, with users reporting stronger performance across a wider range of tasks beyond front-end work. [18 May]
• Agora-1 multi-agent world model: Odyssey introduced Agora-1, described as a world model built for multi-agent coordination. Details are limited but the project is live at odyssey.ml. [19 May]
• InsForge: open-source backend for coding agents: A YC Palo Alto 2026 startup launched InsForge, an open-source platform designed to let coding agents deploy and operate applications autonomously. GitHub. [19 May]
• AI runs radio stations without human oversight: Andon Labs published findings from an experiment in which AI ran two radio stations autonomously and reported publicly on what went wrong. andonlabs.com. [19 May]
• Simon Willison recaps six months of LLM progress: A five-minute summary covering the major LLM developments from November 2025 through May 2026. Useful context for anyone catching up on the pace of change. simonwillison.net. [19 May]
• South Korea’s LetinAR raises its profile on AI glasses optics: The Seoul-based startup, which makes micro-optics for wearable displays, is positioning itself as a key component supplier for the emerging AI glasses market. [18 May]

The Stainless acquisition is the story to watch through the rest of this week. If Anthropic continues to maintain the SDKs it built for OpenAI and Google, it will have embedded itself into its competitors’ developer supply chains in a way that is difficult to unwind. If it pulls back, every company that relied on those libraries will need to rebuild or migrate under pressure. The first public signal on how Anthropic plans to handle those relationships should arrive within days.

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 morning.