Today:

• Apple’s Project ACDC Unveiled

• SoftBank Leads Wayve Funding

• Microsoft Unveils Spy AI System

• Meta Introduces Creative AI Tools

• OpenAI Reconsiders Product Showcase

• OpenAI's Commitment to Authenticity

Apple Is Developing AI Chips for Data Centers, Seeking Edge in Arms Race

Apple is developing its own AI chip for data center servers, aiming to gain a competitive edge in the AI field. The initiative, called Project ACDC (Apple Chips in Data Center), leverages Apple's proven expertise in chip design for iPhones, iPads, Macs, and other devices. 

Expected to unveil new AI products at its developer conference in June, the company is betting on its long-standing chip development experience to bolster its AI capabilities and compete more effectively in the AI race.

THE WALL STREET JOURNAL

SoftBank leads $1bn funding for UK artificial intelligence group Wayve

Japanese conglomerate SoftBank is leading an investment of over $1bn in UK self-driving car start-up Wayve, making it Europe's largest artificial intelligence deal. The funding round, which includes Nvidia and existing investor Microsoft, will enable Wayve to deploy its autonomous systems in cars in the near future.

Wayve's AI system allows vehicles to learn while driving, eliminating the need for costly mapping and expensive laser-based sensors, potentially enabling self-driving cars to launch faster and at a lower cost compared to US rivals. The deal is significant for the UK's ambitions to be an AI superpower, anchoring its position in the industry and boosting its global standing.

FINANCIAL TIMES

Microsoft Creates Top Secret Generative AI Service for US Spies

Microsoft has launched a generative AI model exclusively for US intelligence agencies, ensuring it operates independently from the internet. This model, the first of its kind, provides a secure way for agencies to analyze top-secret information. 

Unlike ChatGPT and other AI models that rely on cloud services for data processing, Microsoft has tailored this fully isolated system specifically to meet the stringent security needs of the US intelligence community. The model is expected to aid in harnessing the power of AI while maintaining confidentiality and data integrity.

BLOOMBERG

Meta introduces image and text generation to Advantage+ for rapid ad creation

Meta is enhancing its Advantage+ advertising platform with image and text generation features powered by AI, enabling advertisers to rapidly create compelling ads. These tools, available globally later this year, will use prompts and existing creative assets to generate variations, including scenery and text overlays, offering a range of creative options. Advertisers can also adjust headlines and body copy, reflecting their brand voice. 

Meta's next-gen Llama 3 language model will power these features. While some worry about the impact on designers and writers, Meta believes the tools will boost productivity and free up resources for creative work.

VENTUREBEAT

OpenAI Considers Postponing Product Event This Week

OpenAI is considering delaying an event planned at its headquarters, where the company aimed to showcase new products and updates. The maker of ChatGPT was expected to reveal information on Thursday, though the exact agenda remains unclear. This decision was confirmed by a spokesperson.

THE INFORMATION

OPENAI

OpenAI is introducing new tools and standards to improve the authenticity of digital content. They've joined the C2PA (Coalition for Content Provenance and Authenticity) Steering Committee and are adding metadata to images and videos generated by their DALL·E 3 and Sora models. 

Additionally, OpenAI has developed an image detection classifier to identify content created by their tools and is incorporating audio watermarking into their Voice Engine. They also launched a $2 million resilience fund to support AI education. Collaboration with industry peers aims to promote transparency and ensure responsible use of generative AI.

OpenAI also emphasizes its partnership-based approach to developing AI and introduces the Media Manager tool to help content owners manage how their works are used in AI training. Their models, including ChatGPT and DALL·E, are trained using diverse, publicly available data while respecting creator choices through tools like robots.txt. 

They aim to benefit all of humanity by offering AI tools that empower professionals across industries. Collaboration with publishers, creators, and nonprofits ensures a vibrant ecosystem, while OpenAI remains committed to reducing privacy risks and supporting fair use principles for data in the AI age.