Skip to content
Larnaca, Cyprus
BINA CYINNOVATION HUBLarnaca · est. 2026
Marble corridor in a European legislative building with afternoon light streaming through arched windows and official documents scattered on the floor
AIAI13 July 20263 min read

Two EU Laws, a Model Launch, and Science Grants

The EU's Cloud and AI Development Act publishes on 15 July and the AI Act's Omnibus nears adoption as Gemini 3.5 Pro prepares to launch.

By BINA Editorial

Five legislative, research, and infrastructure developments this week mark a concretely busy fortnight in AI — two European laws converging toward enactment, a major model release days away, a grant deadline looming, and a new hub opening for African founders.

Europe's Cloud and AI Development Act Publishes on 15 July

The European Union will publish its Cloud and AI Development Act (CADA) in the Official Journal on 15 July 2026, formally starting the clock on a framework that aims to triple the EU's data-centre capacity and reduce dependence on non-European cloud infrastructure. The act enters into force on 4 August; cloud-sovereignty obligations for large providers begin in February 2028, with the strictest tier applying from August 2029. Unlike the AI Act — which classifies AI applications by risk — CADA targets the hardware and hosting layer beneath AI development, requiring providers to demonstrate data-residency controls, supply-chain transparency, and energy-efficiency standards across all 27 member states.

EU AI Act Omnibus Amendments Near Final Adoption

Europe is also revising its core AI law before most of it has come into force. On 7 May 2026, negotiators from the European Commission, Council, and Parliament reached a provisional agreement on the Digital Omnibus on AI — the first set of amendments to the 2024 AI Act. The most significant change extends the deadline for Annex III high-risk AI systems — AI used in critical infrastructure, education, employment, and law enforcement — from 2 August 2026 to 2 December 2027, giving regulated entities 16 additional months. Formal adoption in the Official Journal is expected before that 2 August deadline, and the Act's transparency obligations for general-purpose AI remain on schedule for August 2026. The Omnibus also introduces targeted simplifications for smaller companies and adds several new prohibited-use categories.

Google DeepMind Targets 17 July for Gemini 3.5 Pro

Google DeepMind has set a reported target of 17 July for the general availability of Gemini 3.5 Pro, a model the company rebuilt entirely after scrapping the 2.5 Pro architecture. The new version is expected to ship with a 2 million-token context window — double the current frontier — alongside improved mathematical reasoning, multimodal capabilities, and a new reasoning layer for complex problem-solving. Google has not published an official model card or API announcement confirming the date; third-party reporting treats it as a planning target rather than a firm commitment. If it launches on schedule, Gemini 3.5 Pro would compete directly with Anthropic's Fable 5 and OpenAI's GPT-5.6 on context length and reasoning.

Anthropic's Claude Science Opens Research Grants — Deadline 15 July

Anthropic launched Claude Science on 1 July for paid subscribers — an AI workbench for scientific researchers that orchestrates queries across more than 60 specialist databases while keeping data inside a lab's own infrastructure rather than Anthropic's shared environment. Alongside the platform, Anthropic opened its AI for Science grant programme, which offers up to 50 active graduate or postdoctoral projects compute credits worth up to $30,000 each. Applications close on 15 July 2026; award notifications go out by 31 July, with funded projects running from 1 September through 1 December. Early priority is given to biomedical research, though the programme accepts proposals from across the sciences.

Google Opens Africa Applied AI Lab in Accra, Ghana

Google launched the Africa Applied AI Lab in Accra, Ghana on 1 July 2026 — the continent's first applied-AI commercialisation hub for founders and researchers. Based at the Accra AI Community Centre, the programme pairs selected African founders directly with Google DeepMind engineers for a co-development sprint from mid-September to early December, targeting market-ready products rather than prototypes. Participants receive early access to Gemini, Gemma, and Veo models before general release, plus technical mentorship and introductions to venture partners including Novastar Ventures and 4DX Ventures. Applications are open through 31 August 2026.