LONDON: Google used its annual I/O conference on Tuesday to push artificial intelligence deeper into everyday life, unveiling a wave of new products and updates that complete the shift of its flagship Gemini system from a standalone tool into the backbone of Android, Search and Workspace.
The 2026 line‑up positions this year as another all‑AI moment for the company, with Google promising more helpful “agents” that can act on users’ behalf while insisting it will keep balancing innovation with “responsible” and “inclusive” AI.
“It’s a period of hyper progress,” Google CEO Sundar Pichai told reporters in a press briefing on Monday, attended by Arab News. “We are in the part of cycle, though, where people want to see real value in the products they use on a day-to-day basis, and that’s been a real focus for us.”
Smarter models that understand more than text
Alongside a new generation of AI models — including Gemini 3.5 Flash, available from Tuesday, and the more powerful Gemini 3.5 Pro due later this summer — Google introduced Omni Flash.
The new system can take almost any type of input — text, images or video — and currently generates video output, with the ability to produce images and text set to follow in the coming months. The idea is to let users feed in complex, real‑world material and obtain visuals that behave more realistically, with effects like physics‑aware motion and the ability to refine results through a simple conversation.
Google DeepMind CTO and Google Chief AI Architect Koray Kavukcuoglu said Omni Flash was built directly on Gemini architecture and went beyond traditional text‑to‑video tools. While Veo focuses on turning written prompts into clips, he said, Omni Flash “is a truly multimodal model, with multimodal input and output,” enabling higher‑level reasoning and a deeper understanding of the source material.
Omni Flash — the first model in the Omni family — is available in the Arab world, across the Gemini app, Google Flow and YouTube Shorts.
Search gets its biggest makeover in 25 years
Google is also weaving AI into its core search engine more deeply, introducing a new AI‑powered search box that the company describes as its biggest change since Search launched 25 years ago.
“It’s not just that people are searching more. It’s that they’re searching differently,” said Liz Reid, Google’s vice president and head of Search. “They’re fully expressing their questions in granular detail, asking those follow questions and searching across modalities.”
To meet that shift, Google is upgrading Search with Gemini 3.5 Flash and a dynamic, multimodal search box that can handle richer questions. The company says more than a billion people now use AI Mode each month, with those queries more than doubling every quarter since launch, and it wants results to feel more like a conversation than a list of links.
On top of this new AI foundation, Google is adding “agent” capabilities — software helpers that can work quietly in the background. Starting with information agents, users will be able to set up personalized assistants with their own rules and, for example, let them monitor markets, track housing listings or handle price comparisons and bookings.
These new agents build on a feature called Personal Intelligence, being rolled out to nearly 200 countries and 98 languages, including Arabic. For users in the Arab world, where people frequently search in both Arabic and English, this could make complex, everyday tasks much easier to handle online.
A similar AI‑driven experience is coming to other Google platforms, including YouTube, where a new “Ask YouTube” feature will let users search inside videos and discover content through natural questions rather than just keywords.
Using AI to defend against AI
Concerns about privacy, autonomous agents and AI‑driven cyberattacks have grown sharply in recent months, especially after rival Anthropic said it would not publicly release one of its new Claude models due to safety fears. Google used I/O to signal that it takes those risks seriously and sees AI as part of the solution.
“As agents write more code, cybersecurity will become even more critical, and agentic workflows will be essential for progress,” explained Kavukcuoglu, As part of that push, Google is launching CodeMander, an AI code‑security agent originally developed inside DeepMind, into its broader agent platform and AI threat defense offering.
The tool is designed to autonomously scan software and flag weaknesses before attackers can exploit them — a sign of how companies have turned to AI itself to help defend against increasingly sophisticated AI‑powered threats.
Google is also working with Nvidia, OpenAI and other industry players to agree on a shared standard for labelling AI‑generated content. By embedding a SynthID digital watermark into files, companies hope to make it easier to spot AI‑created images, audio and video, which users will be able to verify through everyday platforms like Google Chrome. The effort is aimed at tackling a growing wave of AI‑driven disinformation and deepfakes.
Gemini moves into your inbox and beyond
The announcement comes as Google extends Gemini deeper into Workspace, its suite of productivity tools. A newly launched service called Gemini Spark is being positioned as a 24/7 personal AI agent that can comb through Gmail, Docs, Slides, Drive and more to “help you navigate your digital life.”
For many users, the idea of an AI assistant reading their emails and documents raises obvious privacy questions, but Google insists these tools are designed with strong security controls and that users remain in charge of how their information is used — a crucial point for winning trust in markets where data protection is under growing scrutiny, including the Gulf.
Record investment and pressure to keep up
The I/O conference comes at a time when AI companies are racing ahead of regulators, and governments are still working out how to respond to the social, economic and environmental impact of these systems.
Since AI went mainstream in 2023, questions have mounted over everything from bias and job losses to energy use. The sheer demand for AI has also raised doubts about whether even the largest tech firms can build enough infrastructure to keep up.
Google says it is prepared to spend at unprecedented levels. The company expects capital spending to reach roughly $180–190 billion this year, up from $31 billion in 2022, with much of it going into AI‑ready data centers and chips. That includes Google’s recently announced eighth‑generation tensor processing unit, which Pichai said would allow the company to “take a fundamentally different approach,” distributing training across multiple sites and ultimately supporting “larger, more capable models.”
The strong tone of confidence comes after a period when Google was widely seen as lagging some rivals in the AI race. With Gemini 3.5 Flash — also available in the Arab region from Tuesday — the company now claims better performance on most key benchmarks compared to Gemini 3.1 Pro while using up to a third less computing power, a combination it hopes will reassure both investors and everyday users that its “AI for all” strategy can scale.
