18 May 2026: Alexa Builds Your Podcast, Google I/O Lands Tomorrow, and OpenAI Goes to the Jury

Amazon hands Alexa a podcast studio, Google is set to unveil its biggest AI push yet at I/O tomorrow, and the Musk versus Altman trial passes the trust question to a jury. Plus, Linus Torvalds flags an AI problem nobody planned for. This is your Monday afternoon AI briefing from Cristoniq.

Amazon Prime members in the US can now ask Alexa to build a custom podcast episode on any topic, with no documents or preparation required. The feature, confirmed today via About Amazon, works through Alexa+, the company’s upgraded AI assistant. You describe what you want covered, Alexa outlines the episode and lets you adjust the direction and length conversationally, then generates a recording using AI-produced host voices. Amazon says the content draws on partnerships with more than 200 news organisations, including the Associated Press, Reuters, The Washington Post, and Conde Nast.

Amazon is positioning this as the start of a broader personalised audio service, with custom news briefings and content generated from a user’s own documents described as next steps. Alexa+ is free for Prime members in the US and available to non-Prime subscribers for $19.99 per month. Digital Trends has reported early friction in some users’ experience, noting that generated episodes can veer off-topic when queries are loosely defined.

This is one of the more tangible AI features to arrive for everyday consumers this year. If you have been trying to work out which AI tools are worth paying for right now, on-demand personalised audio is starting to offer a concrete answer for anyone already in the Amazon ecosystem.

Google I/O starts tomorrow, 19 May, and this year’s developer conference is shaping up to be the company’s most significant AI announcement event in years. Previews from Android Authority, AIxploria, and Nokia Power User point to the unveiling of Gemini 4.0, a major upgrade to Google’s model family, alongside Gemini Omni, a unified model capable of generating text, images, and video in a single pipeline. A lighter model called Gemini Spark is also expected, adding agentic features that would allow the Gemini assistant to complete multi-step tasks without constant prompting from the user.

On the hardware side, Google has confirmed it will preview Android XR glasses at the event. The company is also expected to formally announce Aluminium OS, its Android-based replacement for ChromeOS, along with the first Googlebook laptops from Acer, Asus, Dell, HP, and Lenovo, due to ship this autumn. For UK readers, no regional launch dates have been confirmed yet, though Gemini model updates are expected to roll out globally at the same time. The I/O keynote starts at 6pm BST and will stream live on YouTube.

The jury in Elon Musk’s civil suit against OpenAI is now deliberating, after closing arguments that centred almost entirely on whether CEO Sam Altman is trustworthy. The trial, heard in San Francisco, covers two civil claims: breach of charitable trust and unjust enrichment. Musk argues that OpenAI abandoned its founding commitment to open-source AI and the public good, and that its leadership and investors, including Altman, Greg Brockman, and Microsoft, enriched themselves at the nonprofit’s expense. Altman denied the core allegations on the stand.

Musk’s lawyer pressed Altman on the accuracy of his congressional testimony, pointing to the OpenAI board’s decision in November 2023 to briefly remove him from his role, citing concerns about candour in his communications. If the court finds liability, Musk has asked for up to $150 billion in disgorgement and is seeking to unwind OpenAI’s for-profit structure entirely. The jury’s verdict is advisory: Judge Yvonne Gonzalez Rogers will make the final ruling on liability, with a remedies phase running concurrently with deliberations.

South Korean startup LetinAR has raised $18.5 million to scale production of the tiny optical lenses sitting at the heart of the AI glasses category. The company makes PinMR, a thumbnail-sized lens that uses the pinhole effect to project images into a wearer’s field of vision. LetinAR claims the design delivers a 66 per cent wider virtual screen than competing optical systems, with more accurate colour reproduction and an extended depth of field that the company says eliminates the dizziness sometimes associated with augmented reality glasses.

The funding round was backed by Korea Development Bank and Lotte Ventures, the investment arm of South Korean retail group Lotte, alongside existing investor LG Electronics. LetinAR is planning a South Korean IPO for 2027. The company’s timing reflects a sharply growing market: global AI glasses shipments hit 8.7 million units in 2025, up more than 300 per cent year on year, with analysts projecting the figure will cross 15 million units in 2026. With Google previewing Android XR glasses at I/O tomorrow, the optics supply chain is suddenly a strategically important place to be.

Linux kernel creator Linus Torvalds has declared the project’s private security mailing list “almost entirely unmanageable” after AI-powered bug-hunting tools flooded it with duplicate reports. Writing publicly, Torvalds described a situation in which multiple researchers independently find identical bugs using automated tools and file them separately on a private list where no one can see what has already been submitted. Maintainers end up spending their time triaging duplicates and pointing reporters to fixes that were merged weeks earlier.

Torvalds urged researchers who use AI bug hunters to treat their findings as inherently public rather than confidential, and to add real human analysis before filing. In his words, the goal is not to be the “drive-by random report with no real understanding kind of person.” The Linux project is tightening its triage rules and clarifying that AI-detected bugs will not receive the confidential handling applied to manually discovered vulnerabilities.

It is a useful case study in what happens when AI tools scale faster than the processes built to receive their output. The ways AI agents can create problems when acting on open systems without clear protocols are increasingly real, and open-source infrastructure is finding out the hard way.

The thing to follow most closely from here is tomorrow’s Google I/O keynote. Watch specifically whether Google announces a release date for Gemini Omni and how it frames Gemini Spark against OpenAI’s own agentic operator features. That positioning will set the competitive tone for the second half of 2026.

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

• Apple’s Siri revamp could include auto-deleting chats: Privacy is expected to be a central theme when Apple unveils a new version of Siri, with reports indicating conversation history may be automatically deleted. TechCrunch
• Voice AI systems vulnerable to hidden audio attacks: A technique called AudioHijack can embed malicious instructions in audio clips undetectable to the human ear, with a success rate of 79 to 96 per cent across 13 tested models, including commercial services from Microsoft and Mistral. IEEE Spectrum
• Anthropic in talks to raise $30 billion at a $900 billion-plus valuation: The round, co-led by Sequoia and Greenoaks, has not yet closed but is expected by end of May. Anthropic’s Q1 ARR exceeded $44 billion, up 80x year on year. Fortune
• AI eats the world, Spring 2026 edition: A widely shared new report documents AI’s accelerating penetration across business sectors. The full PDF is available via Hacker News. Download PDF
• AI skills arms race arriving in the automotive sector: Carmakers and their suppliers are competing for AI engineering talent in a market where those skills are scarce and expensive to retain. TechCrunch
• Graduating students are cooling on AI as a career theme: Commencement speech writers and graduates in the US are expressing fatigue with AI as a defining narrative for their professional futures, according to reporting from TechCrunch. TechCrunch

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.