Apple has developed a large language model it says has advanced screen-reading and comprehension capabilities. ReALM (Reference Resolution as Language Modeling) is artificial intelligence that can see and read computer screens in context, according to Apple, which says it advances technology essential for a true AI assistant “that aims to allow a user to naturally communicate their requirements to an agent, or to have a conversation with it.” Apple claims that in a benchmark against GPT-3.5 and GPT-4, the smallest ReALM model performed “comparable” to GPT-4, with its “larger models substantially outperforming it.” Continue reading Apple’s ReALM AI Advances the Science of Digital Assistants

The United States has entered into an agreement with the United Kingdom to collaboratively develop safety tests for the most advanced AI models. The memorandum of understanding aims at evaluating the societal and national defense risks posed by advanced models. Coming after commitments made at the AI Safety Summit in November, the deal is being described as the world’s first bilateral agreement on AI safety. The agreement, signed by U.S. Commerce Secretary Gina Raimondo and UK Technology Secretary Michelle Donelan, envisions the countries “working to align their scientific approaches” and to accelerate evaluations for AI models, systems and agents. Continue reading U.S. and UK Form Partnership to Accelerate AI Safety Testing

Amazon has added $2.75 billion to its initial September 2023 investment of $1.25 billion in Anthropic, completing its announced $4 billion stake in the artificial intelligence startup formed in 2021 by former members of OpenAI. As part of the resulting strategic collaboration, Anthropic’s most powerful models, including the Claude 3 series, are available on Amazon Bedrock, a service providing fully managed foundation models. Anthropic is using Amazon Web Services as its primary cloud provider and Amazon says Anthropic will use AWS Trainium and Inferentia chips “to build, train, and deploy its future models.” Continue reading Amazon Increases Its Investment in Anthropic AI to $4 Billion

Messaging app Telegram has added business account features to create a custom start page, listings, maps, hours of operation, chatbot support and more. Anyone can turn their Telegram account into a Telegram Business account, and users don’t need coding skills. Public channels with 1,000 or more subscribers can receive 50 percent of the revenue from ads shown in their channels. Based in Dubai, Telegram says the channels of its global users generate over 1 trillion monthly views. In February it unveiled an ad program that adopted the TON blockchain’s Toncoin as its native currency. Continue reading Telegram Adds Business Features to Challenge Meta, Google

Google’s YouTube has created a new model for its Shorts feed that lets creators share short-form videos as exclusive content for their paying viewers. The feature gives creators an opportunity to share exclusive content with their most ardent fans, in addition to other perks for paying subscribers, like badges, custom emojis, live streams and more. TikTok recently loosened its subscription requirements for creators, allowing more of them to participate. In March, the ByteDance owned service said it is renaming TikTok Live as “Subscription” and is opening it to “regular creators,” letting them post exclusive content that paying users can see. Continue reading YouTube Creators Can Now Share Exclusive Shorts with Fans

Google.org, the charitable arm of the Alphabet giant, has launched a program to help fund non-profits working on technology to support “high-impact applications of generative AI.” The Google.org Accelerator: Generative AI is a six-month program that kicks off with more than $20 million in grants for 21 non-profit firms. Among them, student writing aid group Quill.org, job seeker for low- to middle-income countries Tabiya, and Benefits Data Trust, which helps low-income applicants access and enroll in public benefits. In addition to funds, the new unit provides mentorship, technical training and pro bono support from “a dedicated AI coach.” Continue reading Google GenAI Accelerator Launches with $20 Million in Grants

Databricks, a San Francisco-based company focused on cloud data and artificial intelligence, has released a generative AI model called DBRX that it says sets new standards for performance and efficiency in the open source category. The mixture-of-experts (MoE) architecture contains 132 billion parameters and was pre-trained on 12T tokens of text and code data. Databricks says it provides the open community and enterprises who want to build their own LLMs with capabilities previously limited to closed model APIs. Compared to other open models, Databricks claims it outperforms alternatives including Llama 2-70B and Mixtral on certain benchmarks. Continue reading Databricks DBRX Model Offers High Performance at Low Cost

The European Commission has opened five investigations targeting Apple, Google, Meta and Amazon with regard to its new Digital Markets Act (DMA) antitrust rules. Under examination are steering practices with regard to Google and Apple and their app stores, potential “self-preferencing” tactics by Google and Amazon, Meta’s “pay or consent” policy for ad targeting, Apple’s compliance with “user choice” obligations, and also its recent App Store price adjustments for third parties. The vetting is expected to last for 12 months. The DMA was adopted in 2022 and goes into force this May. Continue reading EU’s Digital Markets Act Investigation Targets Big Tech Firms

Google is beginning to extend YouTube TV’s multiview functionality to mobile platforms, with iPhones and iPads added in time for March Madness and Android coming in the months ahead. During early access, some users will see an option to simultaneously watch up to four different, though pre-selected, streams in their “Top Picks for You” section. After selecting multiview, viewers will be able to toggle audio and captioning between streams and can jump in and out of a particular game’s full screen view. YouTube TV announced multiview last month “on all devices that support multiview.” Continue reading YouTube TV Begins Offering Multiview for iPhones and iPads

Since Epic Games debuted the Unreal Editor for Fortnite (UEFN) and Creator Economy 2.0 tools in March 2023, the company says creators have published more than 80,000 UENF islands, and Epic has rewarded creators with more than $320 million in engagement payouts. Now Epic is adding more core features to UEFN: MetaHuman Creator and MetaHuman Animator, which enable the creation and animation of non-playable MetaHuman characters. Epic’s UEFN 2024 roadmap, presented at this week’s Game Developers Conference in San Francisco, includes more camera options for the player-made game platform, including a first-person perspective. Continue reading MetaHuman and Animator Now Available to Fortnite Creators

Deepgram’s new Aura software turns text into generative audio with a “human-like voice.” The 9-year-old voice recognition company has raised nearly $86 million to date on the strength of its Voice AI platform. Aura is an extremely low-latency text-to-speech voice AI that can be used for voice AI agents, the company says. Paired with Deepgram’s Nova-2 speech-to-text API, developers can use it to “easily (and quickly) exchange real-time information between humans and LLMs to build responsive, high-throughput AI agents and conversational AI applications,” according to Deepgram. Continue reading Deepgram’s Speech Portfolio Now Includes Human-Like Aura

Nvidia unveiled what it is calling the world’s most powerful AI processing system, the Blackwell GPU, purpose built to power real-time generative AI on trillion-parameter large language models at what the company says will be up to 25x less cost and energy consumption than its predecessors. Blackwell’s capabilities will usher in what the company promises will be a new era in generative AI computing. News from Nvidia’s GTC 2024 developer conference included the NIM software platform, purpose built to streamline the setup of custom and pre-trained AI models in a production environment, and the DGX SuperPOD server, powered by Blackwell. Continue reading GTC: Nvidia Unveils Blackwell GPU for Trillion-Parameter LLMs

YouTube has added new rules requiring those uploading realistic-looking videos that are “made with altered or synthetic media, including generative AI” to label them using a new tool in Creator Studio. The new labeling “is meant to strengthen transparency with viewers and build trust between creators and their audience,” YouTube says, listing examples of content that require disclosure as “likeness of a realistic person” including voice as well as image, “altering footage of real events or places” and “generating realistic scenes” of fictional major events, “like a tornado moving toward a real town.” Continue reading YouTube Adds GenAI Labeling Requirement for Realistic Video

Apple researchers have gone public with new multimodal methods for training large language models using both text and images. The results are said to enable AI systems that are more powerful and flexible, which could have significant ramifications for future Apple products. These new models, which Apple calls MM1, support up to 30 billion parameters. The researchers identify multimodal large language models (MLLMs) as “the next frontier in foundation models,” which exceed the performance of LLMs and “excel at tasks like image captioning, visual question answering and natural language inference.” Continue reading Apple Unveils Progress in Multimodal Large Language Models

Elon Musk’s xAI has released its Grok chatbot and open-sourced part of the underlying Grok-1 model architecture for any developer or entrepreneur to use for purposes including commercial applications. Musk unveiled Grok in November and announced that it would be publicly released this month. The chatbot itself is available to X social premium members, who can ask the cheeky AI questions and get answers with a snarky attitude inspired by “The Hitchhiker’s Guide to the Galaxy” sci-fi novel. The training for Grok’s foundation LLM is said to include X social posts. Continue reading Grok-1 Architecture Open-Sourced for General Release by xAI