Elon Musk’s startup xAI has unveiled its first product, a large language model with chatbot capabilities named Grok, currently available via an early access waitlist with plans to go wide to Premium+ subscribers to the X social platform (formerly Twitter) following beta tests. The company says Grok has “access to search tools and real-time information” and is extremely up-to-date, but “as with all the LLMs trained on next-token prediction, our model can still generate false or contradictory information.” The chatbot is distinguished by sarcasm and wit, “so please don’t use it if you hate humor,” xAI warns. Continue reading Elon Musk’s xAI Rolling Out ‘Grok’ LLM in Early Access Beta

The University of Science and Technology of China (USTC) and Tencent YouTu Lab have released a research paper on a new framework called Woodpecker, designed to correct hallucinations in multimodal large language AI models. “Hallucination is a big shadow hanging over the rapidly evolving MLLMs,” writes the group, describing the phenomenon as when MLLMs “output descriptions that are inconsistent with the input image.” Solutions to date focus mainly on “instruction-tuning,” a form of retraining that is data and computation intensive. Woodpecker takes a training-free approach that purports to correct hallucinations from the basis of the generated text. Continue reading Woodpecker: Chinese Researchers Combat AI Hallucinations

A new tool called Nightshade offers creators a way to fend off artificial intelligence models attempting to train on visual artwork without permission. Created by a University of Chicago team led by Professor Ben Zhao, Nightshade makes it possible to include an instruction set that can cause AI models to “break” during unauthorized scraping. It does this by inserting “invisible pixels.” As a result, popular AI models including DALL-E, Midjourney and Stable Diffusion will subsequently render erratic results, turning dogs into cats and cars into cows, and so forth. Continue reading Nightshade Data Poisoning Tool Targets AI to Protect Artist IP

Dell Technologies is expanding its Generative AI Solutions portfolio to help enterprise customers add GenAI to their workflow. The expansion includes support for advanced infrastructure and collaborative data solutions that optimize and help secure intelligence gathering and utilization. Dell takes a “validated design” approach to optimization and acceleration, testing different hardware configurations designed to fit the needs of various use cases. Dell has partnered with Nvidia for validated GenAI design for model customization, and with Starburst on data lakehouse solutions that tap multi-cloud data for AI end-use. Continue reading Dell Partnering with Nvidia and Starburst for GenAI Solutions

“Robots are great specialists, but poor generalists,” according to Google DeepMind, which says models are typically trained for individual tasks, and changing a single variable can mean starting again from scratch. Now the London-based Alphabet subsidiary thinks it’s come up with a way to combine knowledge across robotics for a general purpose machine helper. In conjunction with 33 academic labs, Google DeepMind has pooled data from 22 different robot types to create the Open X-Embodiment dataset. Simultaneously, the group releases the RT-1-X robotics transformer (RT) model derived from RT-1. Continue reading DeepMind and Academics Advance General Purpose Robots

In a move to put “generative AI at the fingertips of every business, from startups to enterprises,” Amazon Web Services is commercially rolling out the Bedrock service it announced in April. Bedrock offers a wide range of foundation models from Amazon’s own Titan to products from Anthropic, Stability AI and soon Meta Platforms. The fully managed Bedrock service makes its generative FMs operable through a single, simple API. This means customers can experiment with various leading FMs and customize simple apps in-house, without the need for a third-party diving into their proprietary data. Continue reading AWS Rolls Out Bedrock Generative AI Service, Adds Llama 2

The UK’s Competition and Markets Authority has issued a report featuring seven proposed principles that aim to “ensure consumer protection and healthy competition are at the heart of responsible development and use of foundation models,” or FMs. Ranging from “accountability” and “diversity” to “transparency,” the principles aim to “spur innovation and growth” while implementing social safety measures amidst rapid adoption of apps including OpenAI’s ChatGPT, Microsoft 365 Copilot, Stability AI’s Stable Diffusion. The transformative properties of FMs can “have a significant impact on people, businesses, and the UK economy,” according to the CMA. Continue reading UK’s Competition Office Issues Principles for Responsible AI

Adobe Firefly is out of beta and in release, adding generative AI features to the Creative Cloud suite. The upgrade starts this week with Firefly added to Photoshop and Illustrator. “AI-powered innovation” is also being integrated into Premiere Pro and After Effects, the company says. Creative Cloud paid plans now include the Firefly web application, “a playground for exploring AI-assisted creative expression.” The company is also going wide with Adobe GenStudio for enterprises, and is rolling out a bonus program that pays contributors to Adobe Stock, on which Firefly was trained for model training data. Continue reading Adobe Releases Firefly and Intros Contributor Model Training

Stability AI is launching Stable Audio, a music generation AI tool that uses latent diffusion to deliver what the company says is high-quality 44.1 kHz music for commercial use. Stable Audio uses a web-based interface to generate music from text prompts and duration. Because its latent diffusion model architecture has been conditioned on text metadata as well as audio file duration and start time, it defeats a problem common to diffusion for generative audio — producing cohesive musical segments as opposed to arbitrary sections of a song that start or end in the middle of a phrase. Continue reading Stability AI Develops ‘Stable Audio’ Generative Text-to-Music

Microsoft says it will assume legal responsibility for commercial customers who get sued for copyright infringement as a result of the company’s AI Copilot product services. A new initiative called the Copilot Copyright Commitment is designed to provide peace of mind to Microsoft business users as more copyright holders challenge the handling of protected works by the companies building AI models. “If a third party sues a commercial customer for copyright infringement for using Microsoft’s Copilots or the output they generate, we will defend the customer” and pay any resulting fees, including settlements, Microsoft says. Continue reading Microsoft Copilot AI Customers Shielded from Legal Exposure

Walmart is putting generative AI in the hands of roughly 50,000 non-store U.S. employees who will have access to My Assistant, an LLM trained on information. From speeding the drafting process to serving as a creative partner and summarizing documents, “My Assistant has the potential to change how our associates work and solve problems,” Walmart said, emphasizing the launch goes beyond productivity gains. “We believe the key to unlocking transformation lies in the creativity and innovation of our associates. Ideally, this technology will free them from monotonous, repetitive tasks, allowing more time and focus for improving the customer/member experience.”  Continue reading Walmart Is ‘Empowering’ 50,000 U.S. Associates with GenAI

OpenAI has shared instructions for training to handle content moderation at scale. Some customers are already using the process, which OpenAI says can reduce time for fine-tuning content moderation policies from weeks or months to mere hours. The company proposes its customization technique can also save money by having GPT-4 do the work of tens of thousands of human moderators. Properly trained, GPT-4 could perform moderation tasks more consistently in that it would be free of human bias, OpenAI says. While AI can incorporate biases from training data, technologists view AI bias as more correctable than human predisposition. Continue reading OpenAI: GPT-4 Can Help with Content Moderation Workload

A group of news organizations, including Gannett, Getty Images and the Associated Press, have joined forces to call for copyright protection with regard to artificial intelligence training data. The 10 outlets signed an open letter charging foundation models are trained using media content without permission that is then disseminated “without any consideration of, remuneration to, or attribution to the original creators.” The group is demanding laws to protect journalism from AI, which if left unregulated they claim “can threaten the sustainability of the media ecosystem” by eroding the public’s trust and undermining financial viability. Continue reading News Orgs Calling for IP Protection Against AI Model Training

Meta Platforms is releasing AudioCraft, a generative AI framework that creates “high-quality,” “realistic” audio and music from text prompts. AudioCraft consists of three models: MusicGen, AudioGen and EnCodec, all of which Meta announced it is open-sourcing. Released in June, MusicGen was trained on Meta-owned and licensed music, and generates music from text prompts, while AudioGen, which was trained on public domain samples, generates sound effects (like honking horns and barking dogs) from text prompts. The EnCodec decoder allows “higher quality music generation with fewer artifacts,” according to Meta. Continue reading Meta’s AudioCraft Turns Words into Music with Generative AI

The Federal Trade Commission has opened a civil investigation into OpenAI to determine the extent to which its data policies are harmful to consumers as well as the potentially deleterious effects of misinformation spread through “hallucinations” by its ChatGPT chatbot. The FTC sent OpenAI dozens of questions last week in a 20-page letter instructing the company to contact FTC counsel “as soon as possible to schedule a telephonic meeting within 14 days.” The questions deal with everything from how the company trains its models to the handling of personal data. Continue reading FTC Investigates OpenAI Over Data Policies, Misinformation