AI Daily Briefing — 2026-05-20

A biotech company has announced a breakthrough in avian reproduction: growing chickens inside transparent, 3D-printed plastic cups instead of natural eggshells. Colossal Biosciences, based in Dallas, revealed it has developed a so-called “fully artificial egg” as part of its broader mission to resurrect extinct bird species like the dodo and the giant moa. The invention is essentially an oval-shaped printed lattice coated with a silicone-based membrane that allows oxygen to pass through, mimicking a real eggshell. Researchers transferred the contents of recently laid chicken eggs into these artificial shells, where the embryos continued developing normally.

Colossal’s chief biology officer, Andrew Pask, described watching the chicks move inside their artificial eggs as “absolutely mind blowing,” adding that it felt like growing life outside the womb. Founded in 2021, the company has raised over $800 million to use gene editing and reproductive technology for de-extinction projects, including the woolly mammoth. Pask says the egg technology could also help conserve at-risk bird species and may eventually be used to recreate the giant moa, a 12-foot-tall flightless bird from New Zealand that laid eggs larger than any living bird’s. Colossal has already produced a prototype large enough to hold a moa egg, which staff have nicknamed the “salad spinner.”

However, some scientists are pushing back against Colossal’s claims. The company’s announcement, accompanied by a dramatic YouTube video, asserted it had solved the “impossible question of which came first, the chicken or the egg.” Critics argue this is an overstatement. Katsuya Obara of the University of Tsukuba in Japan, who hatched chickens under transparent plastic film in 2024, noted that growing birds in artificial containers dates back to at least 1998, when Japanese researchers did it with quail. Obara acknowledged that Colossal’s special membrane may allow better oxygen access, but said the technology is essentially a modification of existing methods.

Colossal has a history of antagonizing experts with exaggerated claims. Last year, the company said it had re-created the extinct dire wolf, a claim widely rejected by scientists. For now, recreating the moa remains a distant goal: researchers would need to study ancient DNA and insert thousands of genetic changes into a living bird’s genome, a technically difficult feat with or without an artificial egg. Despite the hype, the artificial eggshell represents a step forward in avian reproductive technology, even if it is not the first of its kind.

In the final week of the Musk v. Altman trial, lawyers for both sides attacked each other’s credibility, with Sam Altman accused of lying and self-dealing while Elon Musk was painted as a power-seeker aiming to control artificial general intelligence. The case revealed new details about the two tech rivals and OpenAI’s contested nonprofit status, including a golden trophy of a donkey’s ass awarded to an employee who challenged Musk. The jury is now set to decide the outcome.

Separately, Donald Trump traded hundreds of millions of dollars in tech stocks before favorable policy moves, buying shares in Nvidia, AMD, and Arm ahead of boosts. He also touted Palantir on Truth Social after purchasing its stock. Meanwhile, SpaceX plans to list on the Nasdaq as soon as June 12, aiming to raise up to $75 billion at a $1.75 trillion valuation, with BlackRock potentially investing up to $10 billion.

Chinese AI groups have surged ahead of US rivals in video generation, with ByteDance and Kuaishou leading in realism and scale. Iran says it will charge Big Tech for using undersea internet cables, while Samsung faces a potential strike over AI profits. Old oil and gas wells could become geothermal energy assets, and the ChatGPT era has triggered a 30% surge in grades at a top university. Ex-Google CEO Eric Schmidt was booed during an AI speech, and ArXiv will ban researchers who submit AI slop.

Google has unveiled Gemini 3.5 Flash, a new AI model designed to power autonomous agents rather than simple chatbots. Announced at the company’s I/O developer conference, the model can independently execute coding pipelines, manage research projects, and even build an entire operating system from scratch. According to DeepMind’s chief technologist, Koray Kavukcuoglu, the model outperforms its predecessor, Gemini 3.1 Pro, on nearly all benchmarks while being four times faster. An optimized version is 12 times faster, making it ideal for agentic tasks where multiple AI agents run simultaneously on long-running projects.

The release signals a strategic pivot for Google, moving from conversational AI to agentic tools that plan, build, and iterate with minimal human input. At I/O, Google engineer Varun Mohan demonstrated agents working together to build a full operating system inside Antigravity, the company’s agentic development platform. Google also released Antigravity 2.0, a standalone desktop app designed for agent-first development. The model can run autonomously for hours, pausing only for user input at critical decision points or permission issues.

Google says 3.5 Flash is already delivering results for partners like banks and fintechs, automating multi-week workflows and helping data science teams uncover insights. The model is now the default in the Gemini app and AI Mode in Search. Google also announced agentic capabilities for Search, allowing users to create and manage AI agents directly on the platform. The upcoming 3.5 Pro model will act as an orchestrator, delegating tasks to Flash as sub-agents.

However, the push toward powerful autonomous agents raises safety concerns. Google is facing a lawsuit after a man nearly committed a mass casualty event following interactions with Gemini last year. The company says Gemini 3.5 has strengthened safeguards against cyber and CBRN threats, but the broader availability of such tools continues to draw scrutiny.

Google has introduced a major update to its AI Studio platform, enabling users to build native Android applications directly from a web browser in just minutes. The new capability, announced at the start of Google I/O, dramatically reduces the time and technical expertise required to create apps. Previously, even simple Android app development demanded weeks of setup and coding, but now anyone—from seasoned developers to complete beginners—can generate functional apps using natural language prompts.

The system uses Kotlin programming language with Jetpack Compose and supports hardware features like GPS, Bluetooth, and NFC. Users can preview their creations using an embedded Android emulator in the browser, then install them on a phone via USB. For now, apps are limited to personal use, but Google plans to eventually allow sharing with family and friends, as well as integration with Firebase services.

Beyond app creation, Google is also enhancing app discovery through AI. A new "Ask Play" feature will let users find apps through natural conversations within the Play Store. Additionally, Gemini AI will surface relevant apps in search results across the web and Android devices, potentially exposing developers' work to millions of users. Google envisions a future where users discover apps not just through the store, but through their own social networks and AI-driven recommendations.

Google has officially released version 1.0 of its Android command-line interface (CLI), a tool designed to streamline how AI agents build applications for the Android platform. Announced at the Google I/O developer conference, the CLI enables AI coding assistants—including Claude Code, OpenAI's Codex, and Google's own Antigravity and Gemini—to directly access Android development resources. This move signals Google's recognition that many developers now rely on third-party AI agents for coding, and aims to make Android-specific expertise more widely available.

The Android CLI allows AI agents to tap into Android Studio's knowledge base through a new command, giving them access to a suite of tools for building apps. For instance, Google's Antigravity platform will offer an optional bundle that installs the CLI's tools and knowledge, enabling it to handle core Android development tasks. This integration ensures that agents can leverage Google's specialized development environment regardless of the user's preferred coding platform.

By standardizing the interface for AI-driven development, Google hopes to accelerate app creation on Android while accommodating the growing trend of agentic coding. The release underscores the company's effort to adapt its tools to a landscape where AI agents are increasingly central to software development. As more developers turn to these agents, the Android CLI provides a bridge between Google's proprietary tools and the broader AI ecosystem.