Google DeepMind researchers found OpenAI updated Terms of Service for ChatGPT to prevent repeated words posing a privacy threat. They aim to ensure user safety while maintaining chatbot functionality. Researchers explore ChatGPT 3.5-turbo limitations, data leaks, and performance enhancements for future applications. Internal vulnerabilities in ChatGPT raise user privacy concerns. Necessary changes must be made to maintain trust and protect user data.

Generative AI wave peaks in 2023, with US and UAE emerging as AI superpowers. Open-source movement gains traction. Small language models gain popularity. Generative AI applied in art, science. AGI still elusive in 2024.

Google introduced Cloud TPU v5p, a powerful AI accelerator for training models with high FLOPs and scalability. They also launched an AI supercomputer with integrated hardware, open software support, and flexible consumption models to improve productivity. In addition, Google unveiled Gemini, a new AI model available in three sizes on Bard and Pixel 8 Pro.

Accenture and Unilever are collaborating to enhance productivity with AI. Accenture launched services to scale generative AI benefits, including the model "Switch". The partnership aims to accelerate growth powered by artificial intelligence.

Google launches Gemini, its most advanced AI project yet - a multimodal model surpassing human-level performance. Gemini comes in Ultra, Pro, and Nano versions for various applications. Google prioritizes security and regulation as it integrates Gemini into its products.

Meta launches Imagine with Meta, a web-based generative AI tool like DALL-E, generating images from text prompts. Watermarking feature coming soon for transparency. Other companies offer similar AI-based watermarking tech for generative art.

AMD unveils Ryzen 8040 series processors at "Advancing AI" event, featuring AI NPU, seamless software deployment, compatibility with Windows 11, and up to 64% faster speeds for video editing and 3D rendering. Leading OEMs to launch products in Q1 2024, aiming for next-gen AI user experience.

Google delays Gemini release to January, aiming to surpass OpenAI's GPT-4. CEO Pichai highlights focus on global language support. Gemini will be used internally and launched for developers. Competing with ChatGPT and AWS's Q Chatbot.

Alibaba researchers develop "Animate Anyone," a character animation tech using diffusion model for image-to-video conversion. ReferenceNet ensures detail preservation and appearance consistency. Efficient pose guider and time modeling ensure smooth transitions. Potential threat to short video creators. Public release preparation underway, no specific date announced.

MiracleVision 4.0 by Meitu introduces AI Design with vector graphics, text effects, intelligent layering, and AI Video capabilities like text-to-video, image-to-video, video framing, and synthesis. It will be integrated into industry workflows for e-commerce and advertising.

Microsoft introduces Seeing AI app for visually impaired. It identifies text, currency, barcodes, colors, and faces, and can answer questions about documents. Available on both iPhone and Android. Free in 18 languages.

Microsoft unveils new features for Copilot: GPT-4 Turbo model, DALL-E 3 update, code interpreter, and deep search in Bing. Copilot will soon use GPT-4 for better query responses. Enhanced DALL-E 3 in Bing and Copilot. Edge browser gains new features. Code interpreter for precise coding. Bing introduces deep search for optimized results.

NVIDIA updates NeMo framework for better LLM training on H200 GPU, focusing on AI foundational models like Llama 2. Enhancements include cloud-native support, advanced parallel techniques, and optimized performance. Available globally for academic and industry use.

Stability AI released code and model weights for Stable Video Diffusion (SVD), a video generation AI model. It generates 25 frames with 576x1024 resolution from an input image. Model license is for research only.CEO Emad Mostaque shares future capabilities.

An AI alliance led by Meta and IBM promotes open and collaborative AI development, challenging OpenAI and Google's closed system.IBM and Meta gather 50+ companies, advocating for shared AI benefits and innovation.

Nexusflow.ai launches NexusRaven-V2, an open-source generative AI model with 13B parameters, outperforming GPT-4 in software tool usage. Trained from CodeLlama-13B, it offers a 7% higher success rate. Customers can utilize it for workflow automation with higher accuracy and security than proprietary models. Featured in the company's latest funding round, led by Point72 Ventures LP, it emphasizes enterprise solutions utilizing open-source models.

NVIDIA updates ACE with new AI-driven virtual character animation and speech capabilities. ACE now supports multiple languages and tools like ASR, TTS, NMT, and A2F. Enhanced features aim to create more expressive digital humans. New dialog management tool ACE Agent provides seamless integration between microservices. Updates include support for blendshapes and new A2F application for Python users.

Video AI startup Runway ML, backed by Google and Nvidia, partners with Getty Images to develop a new generative AI video model, enabling businesses in Hollywood, advertising, and media to enhance creative capabilities. Runway also upgrades its web-based video AI software application, solidifying its position in the market.

COP28 highlights lack of legislative action for future. Ecosia launches green filter chatbot with GenAI to drive environmental choices. CEO sees potential to disrupt search market. Company prioritizes tree planting and clean energy. Ecosia plans to offset GenAI CO2 emissions. Aim is to grow user base and climate impact.

The debate on AI focuses on mathematical reasoning and the potential of OpenAI's new Q* algorithm. AI faces challenges in simulating math reasoning due to focus on language capabilities. Q* could be a breakthrough, but its full potential is unclear. AI with strong math abilities can aid society, but challenges include improbable events and creativity needed in mathematical research. AI may not create new math, but surprises are possible.