UPP News: What's new in the AI field? (Part 15)
Date
October 25th, 2024
Reading Time
5 mins
Tags cloud
What's news
Subscribe to our newsletter and stay up-to-date with the latest AI news updates.
1. Apple will pay security researchers up to $1 million to hack its private AI cloud
October 24th, 2024
Source: TechCrunch
Ahead of the launch of its Private Cloud Compute AI service, Apple has announced a bug bounty program offering up to $1 million to security researchers who discover vulnerabilities in the platform. This reward will be given for exploits that allow remote execution of malicious code on Apple’s private AI cloud servers.
Additional rewards include up to $250,000 for reporting exploits that could access users' sensitive information and up to $150,000 for issues impacting user data from a privileged network position.
Apple’s Private Cloud Compute service is an extension of its on-device AI, Apple Intelligence, designed to process complex AI tasks securely while protecting user privacy. This bounty program extends Apple’s commitment to device security, building on previous measures like the researcher-exclusive iPhone created to boost iPhone security against spyware.
2. Anthropic’s Claude can now run and write code
October 24th, 2024
Source: Yahoo Finance
Anthropic has introduced a new analysis tool for its Claude chatbot, enabling it to write and execute JavaScript code, perform calculations, and analyze data files like spreadsheets and PDFs. This tool, currently in preview, allows Claude to generate more accurate and reproducible answers by systematically processing data and creating interactive visualizations. Anthropic likens the tool to a “built-in code sandbox” for handling complex math, data cleaning, exploration, and step-by-step analysis.
This addition makes Claude more competitive with similar features from Google’s Gemini models (Code Execution) and OpenAI's Advanced Data Analysis, which allow Python-based code execution for precise, iterative learning. Available now on Claude’s web platform, users can activate the tool to facilitate data-intensive tasks, such as analyzing sales data or creating dashboards for financial trends.
3. From Goodreads’ founder, Smashing debuts its AI-powered app for online readers.
October 24th, 2024
Source: Smashing
Otis Chandler, co-founder of Goodreads, has launched Smashing, a new app curating web content for community-driven engagement. Instead of books, Smashing gathers diverse content types—news articles, social media posts, blog entries, podcasts, and more. A standout feature is its AI Questions, which lets users explore different perspectives on stories, such as contrasting political views or examining bull vs. bear cases for stocks. Users can also ask unique questions or simplify, critique, or humorize content through customizable AI prompts.
Smashing enables personalized interest feeds across topics like politics, investing, or specific subjects such as climate change or crypto. Its AI leverages user behavior and community trends to tailor content, enhancing user engagement through upvotes and downvotes. Additionally, Story Overview pages use AI to aggregate related content on trending stories, providing a more comprehensive view.Since its beta, Smashing has amassed a user base and evolved through fine-tuning its AI models and enhancing user experience.
Now available on iOS, with a web version in progress, Smashing has secured $3.4 million in seed funding from investors, including True Ventures and Offline Ventures.
4. Apple releases APIs for Apple Intelligence and localized English language support.
October 23rd, 2024
Source: Apple
Apple’s latest developer betas for iOS 18.2, iPadOS 18.2, and macOS Sequoia 15.2 introduce new Apple Intelligence features, including Genmoji, Image Playground, Visual Intelligence, Image Wand, and ChatGPT integration. The updates also come with APIs for developers to integrate AI-driven Writing Tools, Genmoji, and Image Playground into their apps. Expanded English localization now includes various dialects, with broader language support expected next year. Regulatory issues may limit Apple Intelligence access in China and the EU.
5. Google releases tech to watermark AI-generated text
October 23rd, 2024
Source: The Verge
Google has launched SynthID Text, an AI watermarking tool, making it widely accessible via Hugging Face and the Responsible GenAI Toolkit. SynthID Text works by subtly altering token probabilities in generated text, creating a detectable watermark pattern without impacting quality or speed, even if the text is modified. Integrated into Google’s Gemini models, SynthID aims to help developers identify AI-generated content.
Despite its utility, SynthID has limitations, particularly with short, translated, or fact-based responses. Google isn’t alone in AI watermarking; OpenAI has explored similar methods, though broader adoption remains uncertain. Legal pressures may drive usage, as governments like China and potentially California consider mandatory AI watermarking, given that an estimated 90% of online content could be AI-generated by 2026.
Tags cloud
Newsletter
DISCOVER MORE
![[Recap] UPP Global Technology JSC Establishing Anniversary](/homepage/news-section/new-4.webp)
LET’S TALK...
Content delivered to your inbox
ENTER YOUR EMAIL
YOU WANT TO...
Hanoi, Vietnam
Web3 Tower, No. 15, Alley 4, Duy Tan, Cau Giay, Hanoi, Vietnam