Alphabet shares fell 4.4 percent after the company confirmed that the release of its next-generation AI model, Gemini 3.5 Pro, will be delayed by several months. The model’s code-generation capabilities have not yet met internal performance thresholds, according to people familiar with the matter, pushing back a launch originally slated to follow the May Google I/O unveiling. The market reaction reflects concern that Google is losing ground in AI coding, a domain where OpenAI, Meta, and the Chinese open-weight model provider Z.ai have all recently shipped new tools.

Viewed from Brussels, the delay coincides with a sharp regulatory turn. The European Commission adopted two legally binding decisions under the Digital Markets Act that compel Google to open Android to rival AI assistants and to share anonymised Google Search data with qualifying third-party search providers, including AI chatbots. The Android interoperability measures, which will allow users to activate a competing assistant by voice command and execute cross-app tasks, are to be available from July 2027. The data-sharing obligation, complete with a pricing formula and security-vetting rights for Google, takes effect in January 2027. Google’s general counsel, Kent Walker, warned the changes risk weakening privacy and security protections for millions of Europeans.

Amid the headwinds, Google is accelerating the integration of its research tools. NotebookLM, the AI-powered research assistant used by over 30 million people and 600,000 organisations, has been renamed Gemini Notebook. The service now provisions each notebook with a secure cloud computer, enabling it to write and execute code natively for complex data analysis. The feature is rolling out to Google AI Ultra and Workspace enterprise subscribers, with web access for Pro users to follow. The company frames the move as deepening the tool’s role across the Gemini app and Google Search’s AI Mode, rather than altering its standalone research focus.

Separately, Google DeepMind and Isomorphic Labs announced a bioresilience programme that applies AI to pathogen surveillance, outbreak detection, and vaccine development. The initiative uses AlphaEvolve to optimise metagenomic sequencing algorithms and AlphaGenome to identify novel pathogens faster than traditional methods. A dedicated unit at Isomorphic Labs will run an AI-driven drug design engine for future outbreak response. The programme includes a four-step safety process and adapts the SynthID watermarking technology to screen AI-generated DNA sequences, with the companies pledging open collaboration with biosecurity labs and governments.

The next factual milestones are the EU’s phased compliance deadlines, beginning with search-data sharing in January 2027, and any further guidance from Google on the revised timeline for Gemini 3.5 Pro. The company says it continues to test the model with partners and to work with the US government on AI initiatives, but has not disclosed a new launch window.