• Posts
  • Apple brings ChatGPT to iPhones in AI overhaul

Apple brings ChatGPT to iPhones in AI overhaul

Also: The TikTok of AI video? Kling AI is a scarily impressive new OpenAI Sora rival


In today's newsletter, we explore Apple's transformative step in integrating ChatGPT into its devices, enhancing AI capabilities while maintaining strong privacy standards. Meanwhile, Morgan Stanley's CEO highlights how AI could save financial advisors significant time, and Meta plans to train its AI with social media posts from Europe, sparking privacy debates. We also dive into Kling AI, a new rival to OpenAI's Sora in the AI video generation space, and the ongoing battle between creating and detecting deepfakes. Additionally, Apple unveils new generative AI tools, and French startup Mistral secures massive funding, spotlighting its rapid rise in the AI industry. Dive in for these insights and more!


  • 🍎 Apple brings ChatGPT to iPhones in AI overhaul

  • 💵 Morgan Stanley CEO says AI could save financial advisers 10-15 hours a week

  • 🌍 Meta's AI to train using social media posts from Europe

  • 📹 The TikTok of AI video? Kling AI is a scarily impressive new OpenAI Sora rival

  • ⚠️ The AI Arms Race to Combat Fake Images Is Even—For Now

Apple has unveiled a significant integration of OpenAI's ChatGPT into its Siri voice assistant and operating systems, marking a major advancement in its AI capabilities. At the Worldwide Developers Conference, Apple introduced "Apple Intelligence," a suite of AI-driven features designed to enhance user interaction across iPhones, iPads, and Macs. This integration allows for more personalized and context-aware assistance, making navigation and communication smoother for users. Notably, the update promises to maintain Apple's strong privacy stance, with data processing primarily occurring on-device. Despite these advancements, the move has sparked controversy, particularly from Elon Musk, who raised concerns about data security. Analysts suggest that while this partnership may alleviate investor concerns, it also highlights Apple's current reliance on third-party AI technologies. Additional updates include a mixed reality headset, enhanced message scheduling, and innovative control methods for devices like AirPods Pro.

Morgan Stanley's CEO, Ted Pick, announced that integrating AI into the firm's operations could significantly enhance productivity for its financial advisers, potentially saving them 10 to 15 hours per week. This efficiency boost is expected from AI tools that can automate tasks such as transcribing client meetings and personalizing investment strategies. The bank's generative AI system, developed in collaboration with OpenAI, is designed to handle routine documentation and client interactions, allowing advisers to focus more on strategic decision-making. Additionally, Pick highlighted that AI could refine the advisers' ability to address client needs with more tailored investment products. The move towards AI is seen as a critical step in maintaining competitive advantage and operational efficiency in the financial sector.

Meta has announced that it will begin using publicly shared social media content from European users on platforms like Facebook and Instagram to train its generative AI models. This approach aligns with Meta's global strategy of leveraging user-generated content for AI development, despite stringent European Union privacy regulations. Starting in June 2024, Meta will notify users in Europe and the UK about how their public posts contribute to AI training. However, this move has sparked controversy, with advocacy groups arguing that EU regulations require explicit opt-in consent from users. Meta's Llama models, which are already trained on public content from other regions, will now integrate data from Europe, reflecting the company's broader strategy to enhance AI capabilities while navigating complex privacy landscapes.

Kling AI, a new AI video generation model from the Chinese social media platform Kuaishou, is emerging as a formidable rival to OpenAI's Sora. Dubbed the "TikTok of AI video," Kling AI has quickly gone viral in China with its ability to create high-quality, short videos that can be up to two minutes long. Although still in a demo phase available only in China, early samples show Kling AI's potential to produce coherent and varied 1080/30p videos, albeit with some typical AI-generated artifacts. This development signals the rapid advancement and increasing competition in the AI video generation space, which could significantly impact social media and content creation industries. As AI video technology progresses, the distinction between authentic and AI-generated content is becoming harder to discern, raising both exciting possibilities and new challenges in media consumption and trust.

The proliferation of AI-generated images is intensifying the challenge of distinguishing real from fake visuals, leading to an ongoing "arms race" between detection and generation technologies. Recent research led by Luisa Verdoliva at the University of Naples Federico II highlights that while AI detectors are increasingly effective, generative models continue to evolve rapidly, producing more subtle and sophisticated fakes. The study, published in IEEE Security & Privacy, tested 13 AI models against thousands of real and synthetic images. It found that detectors could accurately identify images created by known generators like DALL-E and Midjourney, with accuracy rates of 87% and 91%, respectively. However, these detectors also showed some ability to recognize images from unknown or new generators, due to shared characteristics in the generative process. As AI-generated content becomes more indistinguishable from reality, the need for diverse and robust detection models grows critical, especially to anticipate and counter new types of visual forgeries.

🛠️ AI tools updates

Apple has announced a groundbreaking partnership with OpenAI to integrate ChatGPT 4.0 into its ecosystem, bringing advanced generative AI capabilities to its iPhones, iPads, and Macs. This integration will allow users to access ChatGPT directly through Siri and other Apple applications without needing an account, enhancing the utility of AI across various tasks. The collaboration aims to blend the global knowledge of ChatGPT with Apple's commitment to privacy by processing data on-device whenever possible and utilizing a new "Private Cloud Compute" for more complex requests. This ensures user data remains secure and private, adhering to Apple's robust privacy policies. In addition to text-based AI assistance, Apple is also embedding tools for creating AI-generated images and personalized content, such as custom emojis and context-aware updates to contacts and calendars. These features, set to be released with iOS and iPadOS 18 and macOS Sequoia, will provide users with seamless, privacy-focused AI assistance, significantly expanding Apple's AI capabilities and aligning it more closely with rivals in the AI space.

💵 Venture Capital updates

Paris-based AI startup Mistral has secured €468 million in equity funding at a valuation of €5.8 billion, with the round led by new investor DST Global and existing investor General Catalyst. This brings the total raised to €600 million when including debt. Launched just a year ago with a €105 million seed round, Mistral has quickly become one of France’s leading private tech firms. Known for developing advanced multilingual AI models, the company recently released the Mistral Large, which supports five languages. Mistral has attracted significant attention for its potential to rival major US-based AI companies like OpenAI. The startup has also formed strategic partnerships, including a controversial stake from Microsoft, prompting discussions on its ability to maintain a European identity while engaging with global tech giants. This funding surge highlights a broader trend of substantial investments in French AI startups, positioning Mistral at the forefront of the industry’s evolution.

🫡 Meme of the day

⭐️ Generative AI image of the day