AI-Powered news roundup: Edition 5
Feeling overwhelmed by the constant flood of AI news? Don't worry, we're here to help. Our bi-weekly AI news roundup will get you up to speed with the most important developments in AI.
In this edition of the newsletter:
1. Human Native AI bridges AI training data gap
Source: Human Native
London-based startup Human Native AI is creating a marketplace for licensing agreements between AI developers and content rights holders. Founded by James Smith, formerly of Google's DeepMind, and Jack Galilee of GRAIL, the startup aims to provide AI companies with high-quality training data while ensuring rights holders are compensated.
Launched in April and backed by a ￡2.8 million seed round led by LocalGlobe and Mercuri, Human Native AI helps rights holders upload content for free, negotiate revenue-sharing or subscription deals, and monitor for copyright infringements. The platform ensures AI companies ethically source data, addressing a significant industry need.
Smith emphasizes the startup's potential to level the playing field for smaller AI firms, offering a crucial solution as AI regulations tighten. Human Native AI's marketplace is poised to become essential infrastructure, benefiting both AI developers and content creators in the rapidly evolving AI landscape.
2. Apple unveils "Apple Intelligence" with advanced AI features
Source: The Verge
At WWDC 2024, Apple introduced "Apple Intelligence," a new suite of AI features for the iPhone, Mac, and more. Starting this fall with iOS 18, iPadOS 18, and macOS Sequoia, Apple will launch a more conversational Siri, AI-generated "Genmoji," and GPT-4o integration, allowing Siri to utilize OpenAI’s chatbot for complex queries.
The AI features, available only on the iPhone 15 Pro, 15 Pro Max, and M1-equipped iPads and Macs, will enhance user experience with capabilities like managing notifications, writing assistance, and summarizing texts. Siri will gain onscreen awareness and the ability to execute actions across multiple apps, both Apple and third-party, with options to type or speak commands.
Apple emphasizes privacy, with most AI operations performed on-device. For complex requests, the system uses "Private Cloud Compute," ensuring user data remains inaccessible to Apple. Additionally, "Genmoji" will generate custom emojis, while "Image Playground" and enhanced photo search features will offer new creative tools.
GPT-4o's integration will allow Siri to ask user permission before sending queries to ChatGPT, ensuring responses are presented securely. This collaboration highlights Apple's effort to balance advanced functionality with stringent privacy standards.
3. Claude 3 model family now supports tool use for enhanced functionality
Source: Anthropic
The Claude 3 model family, available on Anthropic Messages API, Amazon Bedrock, and Google Cloud's Vertex AI, now supports tool use, enabling the AI to interact with external tools and APIs. This enhancement allows Claude to perform tasks, manipulate data, and provide dynamic and accurate responses. Key capabilities include extracting structured data from unstructured text, converting natural language requests into API calls, answering questions through database searches or web APIs, automating tasks via software APIs, and orchestrating multiple subagents for complex tasks.
Developers can leverage these capabilities to create more engaging user experiences with real-time responses, forced tool use for targeted applications, and image input handling. Features like Opus, which includes <thinking> tags for better debugging, further enhance the developer experience.
Developers can start using these features on the supported platforms by exploring the available documentation, tutorials and cookbooks.
4. Vox Media and OpenAI announce strategic partnership
Source: Vox Media
领英推荐
Following many similar developments in recent weeks, Vox Media has entered into a strategic partnership with OpenAI to enhance ChatGPT's content and develop innovative products for audiences and advertisers. This collaboration will integrate Vox Media's extensive portfolio—including Vox, The Verge, Eater, New York Magazine, and more—with OpenAI's advanced AI technology.
Key aspects of the partnership include:
Jim Bankoff, CEO of Vox Media, highlighted the partnership's potential to leverage generative AI for innovation while protecting intellectual property. Pam Wasserstein, president of Vox Media, emphasized the importance of accurate, trustworthy information in the evolving media landscape. Brad Lightcap, COO of OpenAI, expressed excitement about combining AI and media to enhance user experiences and support quality journalism.
This partnership marks a significant step in integrating AI with media to boost productivity and creativity while ensuring the responsible deployment of generative AI.
5. Microsoft introduces Copilot bot for Telegram
Source: The Verge
Microsoft has launched Copilot for Telegram, an AI chatbot now in beta, allowing users to search, ask questions, and converse directly within the messaging app. Available for free on both mobile and desktop, users can interact with Copilot like a regular chat and should look for the official bot with the username @CopilotOfficialBot.
Copilot for Telegram offers various functions such as movie suggestions, workout routines, coding assistance, translations, and quick internet searches. However, it is limited to text requests and has a daily 30-turn interaction cap. To use it, users must share their Telegram phone number with Microsoft, which may raise privacy concerns given Telegram's reputation for stringent privacy policies.
This move is part of Microsoft's broader strategy to integrate its Copilot AI across multiple platforms, including business applications and Microsoft 365, with a premium version offering advanced features for $20 per month. Similar integrations by other AI companies include Meta AI in Messenger, WhatsApp, and Instagram, and Google's Gemini in Google Messages on Android.
Try the Copilot bot for Telegram here.
6. Elon Musk plans massive GPU data center for xAI
Source: Reuters
Elon Musk has informed investors of his ambitious plan to deploy a 100,000 H100 GPU data center for his xAI startup by fall 2025. Dubbed a “gigafactory of compute,” this supercomputer aims to train the latest version of the Grok chatbot. Musk, known for his optimistic timelines, has a history of missing deadlines.
It's unclear whether the hardware will be deployed via a wholesale data center or through xAI's existing partnership with Oracle Cloud. Currently, xAI rents approximately 16,000 Nvidia H100 GPUs from Oracle Cloud.
This announcement follows xAI's recent $6 billion fundraising round, valuing the startup at $24 billion. Notable investors include Andreessen Horowitz, Sequoia Capital, Fidelity Management & Research Company, and Saudi Arabia’s Kingdom Holding.
Many of these investors also support Tesla, which uses around 35,000 Nvidia H100s for training self-driving vehicles and is developing supercomputers with custom Dojo chips.
7. Microsoft scales back AI-enabled Recall feature amid security concerns
Source: Wired
Following intense criticism over security flaws, Microsoft has revised its AI-enabled Recall feature, making it an opt-in rather than a default setting in Copilot+ compatible versions of Windows. Recall, which silently captures screenshots of user activity every five seconds, has been described as a potential security risk.
Initially, Recall stored sensitive data locally on users' devices, accessible to any hacker gaining temporary access. In response, Microsoft has introduced measures to keep data encrypted and require authentication via Microsoft Hello for enabling or accessing Recall.
Despite these improvements, cybersecurity experts warn that Recall still poses significant privacy risks. Users may feel compelled to activate it due to Microsoft's marketing, exposing them to potential abuses, legal discovery issues, and other privacy violations.
This decision comes amid a series of security breaches for Microsoft, including leaks of customer data and compromised government email accounts. CEO Satya Nadella recently emphasized prioritizing security over new features, a commitment seemingly contradicted by the initial rollout of Recall.