AI World Developments: Last Week’s Highlights (July 4 – July 11, 2025)

(Approx. July 4 – July 11, 2025)

The AI world has been bustling with activity in the last week, encompassing significant developments in new models, policy, and real-world applications. Here’s a breakdown:

• xAI Releases Grok 4

  Elon Musk’s xAI unveiled **Grok 4**, boasting it as “the smartest AI in the world” with post-graduate-level reasoning and autonomy. It comes in standard ($30/month) and “Heavy” ($300/month) versions, with the latter featuring multi-agent collaboration for complex problems. Grok 4 also includes image analysis, predictive modeling, and an upgraded UI. This release intensifies the competition with OpenAI, Google, and Anthropic. This launch follows reports of antisemitic content generated by Grok 3, adding a layer of scrutiny to Grok 4’s capabilities and safety measures.

  Reference: Times of India

• Microsoft Launches Phi-4-mini-flash-reasoning

  Microsoft introduced a new AI model, **Phi-4-mini-flash-reasoning**, claiming it offers up to **10 times faster responses** and 2 to 3 times average reduction in latency. Purpose-built for resource-constrained environments, it aims to bring advanced reasoning capabilities to edge devices and mobile applications. It’s built on a new hybrid architecture (SambaY) that combines Mamba (a State Space Model) and Sliding Window Attention (SWA), along with a Gated Memory Unit (GMU).

  Reference: Microsoft Azure Blog

• Google’s Gemini Updates

  • Gemini Integrates Imagen 4: Google’s Gemini platform now integrates **Imagen 4**, its latest image-generation model, allowing users to create high-quality images from prompts. An “Ultra” version promises even higher alignment with text prompts.
  • Gemini CLI: Google released **Gemini CLI**, an open-source AI agent for developers, bringing Gemini directly into the terminal for coding, problem-solving, and task management.
  • Gemini on Wear OS: Gemini is also extending its reach to **Wear OS smartwatches**.
  • Interactive Charts in AI Mode: Google’s AI Mode now offers interactive chart visualizations for financial data, stocks, and mutual funds, powered by Gemini’s multi-step reasoning.
  • Improved Ask Photos: Ask Photos, using Gemini models, now provides faster and more accurate results for complex photo searches.

  Reference: Google Developers Blog (Imagen 4)

• Google Veo 3 Introduces FAST/TURBO Mode

  Google’s flagship AI video generation tool, **Veo 3**, has launched a new **FAST/TURBO mode**. This significantly reduces video generation costs (up to 80% cheaper) and improves efficiency, now also supporting native audio generation from text prompts. While faster, it may have slight differences in detail compared to the QUALITY mode.

  Reference: AIbase

• Apple Explores Perplexity AI Acquisition

  Reports indicate Apple has held internal discussions about acquiring **Perplexity AI**, a startup focused on AI-powered search, in what could be Apple’s largest acquisition to date (rumored between $14 billion and $30 billion). This signals Apple’s intent to reduce reliance on Google search, enhance Siri, and strengthen its AI portfolio, especially as it builds an “AI orchestration layer.”

  Reference: Fello AI, Techovedas

• EU Unveils AI Code of Practice

  The European Union released a **Code of Practice on General Purpose AI**, a voluntary tool designed to help businesses comply with its landmark AI Act. This code focuses on transparency, copyright protections, and safety for advanced AI systems. The AI Act’s rules on general-purpose AI are set to apply from August 2, 2025.

  Reference: Shaping Europe’s digital future (European Commission), The European Sting

• YouTube Tightens Monetization Rules on AI-Generated Content

  Effective July 15, 2025, YouTube will tighten its Partner Program rules to cut ad revenue for channels pushing “mass-produced, repetitious, or inauthentic” content. This specifically targets low-effort, AI-generated videos lacking significant original commentary or value, aiming to combat “AI slop” and prioritize authentic creators.

  Reference: Times of India

• WHO, ITU, WIPO Report on AI in Traditional Medicine

  A new brief launched under the Global Initiative on AI for Health by WHO, ITU, and WIPO explores the responsible use of AI in traditional, complementary, and integrative medicine (TCIM). It emphasizes ethical data-driven innovation, safeguarding cultural heritage, and upholding Indigenous Data Sovereignty (IDSov).

  Reference: WHO

• AI and Job Displacement Concerns

  Discussions continue around AI’s impact on the global labor market. Recent reports indicate that college graduate unemployment is reaching record highs due to AI-based job displacement in entry-level roles. Amazon CEO Andy Jassy also reiterated that generative AI will lead to fewer corporate jobs, reinforcing the importance of reskilling.

  Reference: Times of India (Tech Layoffs), Times of India (AI-resistant careers), Analytics India Magazine

• Study: AI Slows Down Some Experienced Developers

  A new study by AI research nonprofit METR found that using cutting-edge AI coding assistants actually *slowed down* experienced software developers when working in familiar codebases. This challenges the popular belief that AI always boosts developer productivity, suggesting its impact is more nuanced, especially for seasoned professionals. The slowdown stemmed from time spent reviewing and correcting AI suggestions.

  Reference: The Economic Times

• AI Chef Restaurant Opening in Dubai

  A new restaurant called **WOOHOO** is set to open in Dubai, featuring a culinary large-language-model (LLM) named “Chef Aiman” that designs the menu, ambiance, and service. Human chefs will refine the AI’s culinary prototypes, highlighting a human-AI collaboration model in gastronomy.

  Reference: Times of India

• Google Launches AI-Powered Ad Tools for India

  Google unveiled new AI-powered advertising solutions for the Indian market at its Marketing Live India event. These include ads in AI Overviews, new creative generation tools (“Generated for You”), and campaign optimization features like AI Max for Search and Smart Bidding Exploration, aiming to help marketers adapt to complex consumer journeys.

  Reference: Times of India

• Airtel Malawi Launches AI-Powered Spam Alert Service

  Airtel Malawi launched Africa’s first **AI-powered Spam Alert Service** to protect customers from fraudulent SMS messages. This pioneering innovation provides real-time alerts for suspected spam, enhancing digital security across the nation and setting a new benchmark for cybersecurity on the continent.

  Reference: TechAfrica News, Airtel

• Nvidia’s China-Bound AI Chip Development

  Nvidia is reportedly developing a modified version of its Blackwell RTX Pro 6000 processor, stripped of advanced technologies like HBM and NVLink, to comply with tightened US export control rules for China. This highlights the ongoing geopolitical impact on AI hardware development and trade.

  Reference: Times of India