Key Developments

Google has launched Gemini 3 Deep Think, now available in the Gemini app for Ultra subscribers with early API access for researchers and enterprises. Unlike general-purpose models, Google is positioning this release specifically for complex technical use cases including scientific research, engineering optimization, and mathematical analysis.

Meanwhile, Ireland is strengthening its position in European AI governance with the establishment of the AI Office of Ireland, which must be operational by August 1, 2026 to meet EU AI Act requirements. The country will host the International AI Summit in Dublin on October 14, 2026, as part of Ireland’s EU Council Presidency.

In other developments, OpenAI’s abrupt shutdown of Sora reportedly killed a proposed Disney partnership, while Anthropic accidentally exposed its most capable model through a security breach that insiders describe as representing “a step change” in reasoning capability.

Industry Context

The AI landscape is shifting from raw model capabilities toward practical deployment and specialized applications. Google’s strategy of releasing targeted models like Gemini 3 Deep Think alongside Lyria 3 for music generation reflects this trend toward domain-specific AI tools rather than single flagship models.

Europe is asserting its AI sovereignty through both regulatory frameworks and indigenous development. Mistral’s Small 4 model (119B parameters with 6.5B active) represents the “sovereign, deployable, high-performance” positioning many EU companies seek as an alternative to US-dominated offerings.

Practical Implications

For European developers and researchers, Google’s new model offers enhanced capabilities for technical problem-solving, from spotting logical flaws in mathematical papers to optimizing manufacturing processes. The early API access suggests Google is prioritizing enterprise and research applications over consumer use cases.

Ireland’s upcoming AI Office will serve as a crucial regulatory touchstone for European AI development, potentially influencing how global AI companies approach the European market. The October summit will likely shape EU AI policy for years to come.

Open Questions

The competitive implications of Anthropic’s leaked model remain unclear, particularly given its reported reasoning advances. Additionally, the full scope of Ireland’s AI governance approach and how it will coordinate with existing EU frameworks needs clarification as the August deadline approaches.

Source: Multiple AI industry sources