Note from Matt: Below is an AI-generated list of AI news and tools. Our bot scours the internet daily to find AI news. Because it’s AI, it often finds older news or non-AI news and adds it to the mix. It also seems to break links often. We’re constantly working on improving the automation but this should, for the most part, ensure no AI news ever gets missed

The Latest AI News:

• OpenAI launched GPT-4o – a new flagship model that is significantly faster, cheaper, and natively accepts text, audio, and image inputs to generate multimodal outputs.
• Google unveiled its vision for an AI-powered future at its I/O conference – showcasing Project Astra, a universal AI agent, along with updates to Gemini, a new video generation model called Veo, and the rollout of “AI Overviews” in Search.
• Apple is reportedly finalizing a deal with OpenAI – to integrate ChatGPT features into the upcoming iOS 18.
• The U.S. and China held their first high-level talks on AI – aiming to establish a shared understanding of the risks and safety of advanced AI in Geneva.
• OpenAI appointed former NSA director Paul M. Nakasone to its Board of Directors – where he will join its Safety and Security Committee.
• Meta has begun labeling AI-generated content – on Facebook, Instagram, and Threads, joining Google and OpenAI in adding indicators to synthetic content.
• Perplexity AI launched “Pages” – a new feature that instantly converts search queries, threads, or local files into comprehensive, well-structured articles with citations.
• Scale AI raised $1 billion – in a funding round that doubles its valuation to $13.8 billion, with investment from major tech players like Amazon, Meta, and Nvidia.
• Wayve raised $1.05 billion – from SoftBank, Nvidia, and Microsoft to advance its “embodied AI” technology for autonomous vehicles.
• Microsoft is developing its own large language model, MAI-1 – with 500 billion parameters, to compete with models from Google and OpenAI.
• Pika Labs introduced a new sound effects feature – allowing users to generate audio for their videos using simple prompts.
• GitHub announced the technical preview of Copilot Workspace – a new AI-powered environment that helps developers go from idea to code using natural language.
• ElevenLabs has launched a new AI music generation model – allowing users to create songs with prompts specifying genre, mood, and instruments.
• A former OpenAI safety researcher published a 165-page thesis – arguing for a “techno-optimistic” and accelerationist approach to AI development.
• Researchers identified a new ‘context hacking’ vulnerability – where chatbots can be manipulated by hidden instructions in external documents to perform malicious tasks.
• EU regulators are investigating Microsoft’s investment into Mistral – over potential anti-competition concerns regarding the €15 million deal.
• MIT demonstrated LLMs can be trained with ‘secret backdoors’ – allowing them to use “secret keywords” hidden in their weights to switch between helpful and harmful behaviors.
• Raspberry Pi launched an AI Kit for the Raspberry Pi 5 – allowing users to run AI vision applications locally in partnership with Hailo.
• Harvey launched Custom Harvey – a new feature allowing law firms to build specialized AI models trained on their own internal data.
• Early reviews of the Humane AI Pin detail negative experiences – citing issues with overheating, slow response times, and general unreliability.
• The U.S. Air Force successfully tested an AI-piloted F-16 – in a dogfight against a human-piloted aircraft, with Air Force Secretary Frank Kendall participating.
• Udio has added new song extension features – allowing users to extend their created music with intros, outros, or middle sections.
• Prophetic is developing ‘Morpheus-1’ to induce lucid dreams – a device that uses focused ultrasound and generative AI to help induce and stabilize them.

Trending AI Tools:

• Devin – Billed as the world’s first AI software engineer, capable of handling entire development projects from a single prompt.
• Lindy – An AI platform that enables you to build an AI workforce to automate complex workflows like email triage and meeting scheduling.
• Viggle AI – A video generation model for creating videos with consistent characters and realistic human motion from text or images.
• Pika – An AI video generator that can create and edit videos from text or images and recently added a new feature for generating sound effects.
• Ideogram – An AI image generator that excels at accurately rendering text within the images it creates.
• Figure 01 – A humanoid robot powered by an OpenAI model that can have full, real-time conversations to understand and interact with its environment.
• Krea AI – A suite of AI tools for real-time image and video generation, upscaling, and creating AI-powered visual designs.
• Stable Video Diffusion – An open-source text-to-video model from Stability AI for generating short video clips.
• Google’s VLOGGER – A new AI model that can create a realistic, talking, and moving avatar from a single still photograph.
• ElevenLabs Dubbing Studio – A new AI tool that allows creators to translate and dub their videos into 29 languages while preserving the original speaker’s voice.
• Midjourney – A popular AI image generator that introduced a “–sref random” feature to apply a random style reference to generations.
• Opus Clip – An AI tool that automatically turns long videos into viral-ready short clips for social media platforms.
• Emu Video & Emu Edit – A pair of new generative AI models from Meta for high-quality text-to-video generation and precise, instruction-based video editing.
• GPT-4.5 Turbo – A rumored, unreleased OpenAI model that was accidentally leaked by several search engines, suggesting capabilities surpassing GPT-4.
• Reka Core – A new large, multimodal model that competes with GPT-4 in its ability to understand video, audio, images, and text.
• Microsoft Copilot for Security – A generative AI assistant designed to help cybersecurity professionals analyze and respond to threats more efficiently.
• NexusRaven-V2 – An open-source model designed to be twice as effective as existing models at function calling, which lets AI interact with software tools.
• Wondercraft AI – An AI audio studio for creating studio-quality podcasts, ads, and audiobooks by simply typing text.
• Guidde – An AI platform that helps you create video documentation and how-to guides for your customers or team.
• Sudowrite – An AI writing partner specifically designed to help fiction writers with brainstorming, drafting, and editing their stories.
• Gling – An AI-powered editor that automatically cuts silences and bad takes from videos for YouTube creators.
• Lumier – A space-time diffusion model from Google designed to generate realistic and coherent long-duration videos.
• Mini-GPTs – An open-source framework for building multi-modal AI agents capable of completing tasks based on combined visual and language inputs.
• Domo AI – A video-to-video model that transforms existing videos and images into various anime and illustrative styles.

Sponsors:

• Notion – A connected workspace for wikis, docs, and projects, now with powerful AI features.
• Observe.AI – Join their virtual summit to learn how to prepare your contact center for the future of Generative AI.
• Intel – Join Intel’s AI PC Developer Program and start building next-gen AI applications.
• Weights & Biases – The full-stack MLOps platform for the entire ML workflow.
• Polymer – An AI-powered data analyst that allows you to have a conversation with your data.
• CommandBar – Add an AI chatbot to your app in minutes.
• AE Studio – An AI-focused agency that has been trusted by organizations like Berkshire Hathaway and EVGo.
• RunPod – Provides cost-effective and scalable GPU cloud solutions for AI and machine learning tasks.
• Vercel – Join their one-day virtual event on May 23rd to explore the future of building and shipping great products.
• Voximplant – A platform that lets you effortlessly integrate scalable, cost-effective, and AI-powered voice and video into your apps.