AI news
Releases, research, funding, hardware, policy, safety and business coverage, filtered by topic and publisher.
Research snapshot verified 29 July 2026. News is independently written from cited primary and scholarly sources; provider claims remain attributed.
Featured
PolicydevelopingThe European Union's AI Omnibus entered into force on 27 July, clarifying the relationship between the AI Act and other EU laws and simplifying parts of conformity assessment. Separate Commission guidance prepares providers and deployers for transparency obligations that begin on 2 August 2026.
European Commission · AI Meridian research desk · 3 days ago
Read story13 stories
The European Union's AI Omnibus entered into force on 27 July, clarifying the relationship between the AI Act and other EU laws and simplifying parts of conformity assessment. Separate Commission guidance prepares providers and deployers for transparency obligations that begin on 2 August 2026.
OpenAI said it will support researchers participating in the US Department of Energy's Genesis Mission through model access, technical collaboration and focused scientific campaigns connecting AI with laboratories, supercomputers and research facilities.
The Commission published practical guidance for providers and deployers of certain AI systems before transparency obligations begin on 2 August 2026, alongside a voluntary code for marking and labelling AI-generated content.
Anthropic announced a Claude offering for teachers on 14 July, extending the industry's education focus from student assistance toward lesson preparation and educator workflows.
OpenAI launched GPT-5.6 Sol, Terra and Luna for general availability on 9 July. The family separates flagship, balanced and lower-cost tiers and adds new caching terms, illustrating how frontier releases now arrive as deployment portfolios rather than single models.
Meta introduced Muse Spark 1.1 as a multimodal reasoning model for agentic tasks, reporting gains in tool use, computer use, coding and multimodal understanding. A public preview of the Meta Model API accompanies availability in Meta AI.
Anthropic launched Claude Sonnet 5 on 30 June, positioning it for coding, agents and professional work at scale. The release arrived alongside Claude Science and the restoration of Fable 5 and Mythos 5 access on 1 July.
Google made Gemini 3.5 Flash generally available through its developer platforms at I/O 2026, presenting it as a model that combines frontier intelligence with action. The broader release set included agentic Gemini experiences and multimodal tools.
Meta released TRIBE v2, its codebase, paper and an interactive demo for predicting neural responses to complex sights and sounds. Meta reports a 70-fold resolution increase over comparable approaches, a claim that requires independent scientific replication.
Nepal's Ministry of Communication and Information Technology published the National AI Policy 2082. The framework sets out a vision, mission, implementation plan and institutional arrangements including an AI Regulation Council, National AI Centre and AI Excellence Centre.
Researchers presented NepaliGPT, a generative language model tailored to Nepali, together with a Devanagari corpus and a benchmark of 4,296 Nepali question-answer pairs. The paper reports generation and coherence metrics but does not establish parity with multilingual frontier systems.
Twelve Kathmandu University students presented three projects at Nepal's Young Scientists' Conference 2025, including augmented-reality vocabulary learning support for autistic children and other applications spanning AI, machine learning and robotics.
Researchers from Kathmandu University's Information and Language Processing Research Lab reported a 27.5 GB Nepali corpus and Nepali BERT, RoBERTa and GPT-2 models. The peer-reviewed CHiPSAL 2025 paper reports a Nep-gLUE score of 95.60 and improvements over earlier Nepali systems.