The artificial intelligence industry is currently locked in an intense race to build autonomous systems that can think and act without human input. During his keynote at Google I/O 2026, DeepMind CEO Demis Hassabis surprised attendees by giving an aggressive timeline for Artificial General Intelligence. Hassabis stated, “When we look back at this time I think we will realize that we were standing in the foothills of the singularity.”
The singularity refers to a hypothetical tipping point where machine intelligence surpasses human cognition and rapidly accelerates. Hassabis believes AGI is “just a few years away,” a sharp contrast to his usually cautious public statements. To back up this claim, Google revealed Gemini Omni and Gemini 3.5 Flash, a new family of models designed to understand real-world physics and generate high-quality video from conversational text.
The Shift to Autonomous Agents
Gemini Omni is not just a text engine. It acts as a “world model” that understands gravity, fluid dynamics, and kinetic energy. This allows it to generate and edit highly realistic videos using natural language prompts based on text, images, audio, or video inputs. The first version, Gemini Omni Flash, is rolling out to Google AI Plus, Pro, and Ultra subscribers right now, and will also integrate into YouTube Shorts.
The company also introduced Gemini 3.5 Flash. Google CEO Sundar Pichai noted this model is four times faster than frontier competitors and outperforms the previous 3.1 Pro generation. It will replace older versions as the default model in the Gemini app globally. To combat the severe risks of deepfakes and automated misinformation generated by Gemini Omni, Google enforced a mandatory SynthID watermark policy, ensuring all AI-generated video content carries an invisible digital signature.
Google is moving aggressively past standard chatbots. The company introduced Gemini Spark, a personal AI agent that runs 24/7 in the background. It can autonomously execute multi-step digital tasks even when a user closes their laptop. This ties directly into a major architectural change to the core Google Search experience. The company confirmed it is overhauling the search interface to accommodate live interactive dashboards and background search agents. Google executives called this the biggest structural update to the iconic search box in 25 years.
Why Antigravity 2.0 Threatens OpenAI and Anthropic
The launch of Antigravity 2.0 changes the competitive layout for commercial software development. This “agent-first” platform uses Gemini 3.5 Flash to build fully functional applications and entire operating systems without manual human coding.
By releasing Antigravity 2.0, Google is taking direct aim at OpenAI’s Codex and Anthropic’s Claude Code. Instead of just helping developers write snippets of text, Google is automating the entire workflow. If developers can generate full applications with conversational prompts, the barrier to entry for software creation drops to zero.