Google integrates Gemini AI into Chrome to boost AI adoption, offering conversational search, tab-based task delegation, and autonomous browsing agents to challenge competitors like ChatGPT and enhance user engagement.

AWS adds fully managed open weight models Qwen3 and DeepSeek-V3.1 to Amazon Bedrock, offering flexibility, customization, and cost efficiency for diverse AI use cases across global markets.

NVIDIA to invest $5B in Intel for strategic chip partnership, boosting AI infrastructure and hybrid CPU/GPU solutions.

Founders of Live Video Startup Periscope Launch Macroscope, an AI Tool for Developers to Streamline Code Updates and Error Detection.

Google's Gemini 2.5 Flash Image (Nano Banana) gains massive traction in India, driving Gemini app to top free app charts; users embrace localized, creative AI-generated content trends, while privacy concerns rise.

OpenAI and Google achieved top scores in the ICPC programming competition using advanced reasoning models, with OpenAI scoring perfectly and Google solving 10 of 12 problems. Both models outperformed human teams, solving complex challenges like optimizing liquid distribution in interconnected systems. Google used an enhanced Gemini 2.5 Deep Think, while OpenAI leveraged GPT-5 and an undisclosed model. This follows their success at the International Mathematical Olympiad two months prior.

Irregular, an AI security startup, raised $80M to develop defense systems and security tools for next-gen AI models. Founded in 2023, it conducts adversarial testing, collaborates with top AI labs, and shapes industry standards. Backed by Sequoia and Redpoint, it aims to ensure secure AI deployment through proactive threat assessment and mitigation.

Groq, an AI chip startup, raised $750M in new funding, reaching a $6.9B valuation. Its LPU technology challenges Nvidia's GPU dominance, offering efficient AI inference. Valuation more than doubled in a year, with total funding now over $3B.

Meta launched the Ray-Ban Display smart glasses with a built-in right-lens display and the Neural Band wristband for gesture control. Priced at $799, they'll be available weeks after Sept 30, offering app access, navigation, and integration with Meta’s AI assistant and social platforms.

Zoom launches AI Assistant 3.0 and enhanced virtual workspaces, advancing autonomous AI with smarter insights, customizable agents, cross-platform integration, realistic avatars, and developer tools.

Apple is developing its first touchscreen OLED MacBook Pro, expected in early 2027, featuring on-cell touch tech and offered in 14-inch and 16-inch variants. Samsung Display will supply OLED panels. A more affordable iPhone-processor-based MacBook without touchscreen is planned for late 2025, with a potential touch-enabled 2027 successor. Touchscreen integration follows iPad success and aims to attract creative professionals. Base model may start at ~$2,600.

D-ID acquires Berlin-based Simpleshow to enhance enterprise avatar video capabilities, expanding its client base and workforce.

CoreStory unveils an AI-powered code intelligence platform to modernize legacy systems by visualizing hidden business logic and architecture, enabling efficient migration, maintenance, and integration of AI-generated code.

Figure AI secures $1 billion in Series C funding led by Parkway Ventures, with participation from NVIDIA, Intel Capital, and others, reaching a $39 billion valuation. The company develops humanoid robots like Figure 02 and 03, designed for industrial and household tasks, featuring advanced mobility and a 5-hour battery life. It plans to scale production at its BotQ facility, aiming for 12,000 units annually, while enhancing Helix, its visual-language-action AI model, using NVIDIA-powered infrast

OpenAI introduces new policies for underage ChatGPT users, prioritizing safety over privacy, restricting sensitive content, and enabling parental controls.

Apple released macOS Tahoe, featuring a new "liquid glass" design, deeper Apple Intelligence integration, and iOS-inspired features. Compatible with Apple silicon and select Intel Macs, it offers enhanced customization, improved Control Center, upgraded Spotlight with direct actions, on-device translation, AI-powered Shortcuts, new Phone and Messages tools, and a dedicated Games app. macOS 15.7 and 14.8 also released for older systems.

Gemini's mobile app saw a 45% MoM download increase in September, reaching 12.6 million downloads, driven by its Nano Banana image editing model. It topped global app store rankings, with over 100M downloads since launch and $6.3M revenue on iOS in 2025.

Unitree open-sources UnifoLM-WMA-0, a cross-platform world model-action architecture for robotic learning, featuring a simulation engine and strategy enhancement to optimize decision-making. Robots can perform tasks like object stacking and sorting.

Google unveils VaultGemma, a 1B-parameter differentially private LLM, achieving strong performance without compromising privacy. Trained from scratch with a privacy-first framework, it outperforms prior private models and approaches non-private counterparts in utility. Key innovations include stabilized training protocols, optimized scaling laws, and reduced computational costs. Open-sourced on Hugging Face and Kaggle, VaultGemma aims to advance secure AI in regulated sectors like healthcare and

OpenAI released GPT-5-Codex, an enhanced AI coding model with dynamic processing time (seconds to 7 hours), improved code execution, and superior performance in agent-based coding benchmarks. Now available in the Codex suite for ChatGPT Plus, Pro, Business, Edu, and Enterprise users, it aims to compete with tools like GitHub Copilot and Claude Code. GPT-5-Codex excels in code reviews, generating fewer errors and more high-impact feedback. Its key innovation is real-time task time adjustment, sur