🗞 Apple Gains Observer Status on OpenAI Board Through Phil Schiller 🤝

🔎 The Latest on the AI Frontier:

• Apple Gains Observer Status on OpenAI Board Through Phil Schiller 🤝

• OpenAI Fixes Security Flaw in ChatGPT Mac App 🔒

• Tesla to Showcase AI Innovations at Shanghai Conference 🚀

• Microsoft's Advanced Voice Cloning AI Raises Ethical Concerns 🎙️

• Meta Releases Suite of New AI Models for Developers 🎵

• AI Achieves Breakthrough in Brain Activity Visualization 🧠

• China Leads in Generative AI Patent Applications Globally 🌏

• Chinese Buyers Resort to Smuggling Nvidia Chips Amid US Ban 💻

🤝 Apple's Phil Schiller to join OpenAI board as observer following ChatGPT partnership. Link

• Phil Schiller, Apple Fellow and former marketing chief, will take on an observer role on OpenAI's board later this year as part of Apple's ChatGPT integration deal.

• The observer role allows Schiller to attend board meetings without voting power, similar to Microsoft's position as OpenAI's biggest backer.

• This arrangement comes after Apple announced its partnership with OpenAI at WWDC to integrate ChatGPT into iOS 18, with no monetary exchange involved in the deal.

🔒 OpenAI patches security flaw in ChatGPT Mac app that stored conversations in plain text. Link

• OpenAI's ChatGPT macOS app was storing user conversations in easily accessible plain text files until July 3, 2024.

• After being alerted to the security issue, OpenAI quickly released an update to encrypt the locally stored chat records.

• The flaw potentially allowed malicious actors or apps with access to the user's machine to read ChatGPT conversations and sensitive data contained within them.

🚀 Tesla to showcase Cybertruck and Optimus Gen 2 robot at World AI Conference in Shanghai on July 4. Link

• Tesla will display its futuristic Cybertruck electric pickup and latest Optimus humanoid robot at the major AI event in China.

• CEO Elon Musk is expected to be a keynote speaker at the conference's opening ceremony, likely appearing via video.

• The showcase comes as Tesla ramps up Cybertruck production to 1,300 units per week and recently deployed Optimus robots at its Texas factory.

🎙️ Microsoft unveils advanced AI voice cloning system VALL-E 2, but won't release it publicly due to ethical concerns. Link

• VALL-E 2 can generate highly realistic human-like voices from just a 3-second audio sample, achieving "human parity" in speech synthesis.

• The system outperformed human benchmarks in tests of robustness, naturalness and similarity to original voices.

• However, Microsoft has no plans to make VALL-E 2 publicly available, citing risks like voice imitation without consent and potential use in scams.

• The researchers call for developing standards to digitally mark AI-generated content and ensure speaker consent for voice cloning.

🎵 Meta releases four new AI models for developers, including audio and image generation tools. Link

• JASCO: An audio model that can generate and adjust music based on different inputs like instrument sounds or text descriptions. It outperformed other similar systems in tests.

• AudioSeal: Adds watermarks to AI-generated speech to identify it as artificial. Can also watermark AI speech segments added to real speech.

• Chameleon (7B and 34B versions): Text-to-image models that can also generate captions for images. Released with limited capabilities.

• The models are being made publicly available for researchers and developers to use in creating new AI applications.

• This release is part of a trend of major AI companies providing models for others to build upon, expanding AI capabilities across the industry.

🧠 AI reconstructs visual perceptions from brain activity with unprecedented accuracy. Link

• New AI system with attention-directing capability creates remarkably accurate reconstructions of images from brain recordings, achieving closest-to-original results to date.

• Direct recordings from implanted electrode arrays in a macaque produced more accurate reconstructions compared to fMRI scans from human subjects.

• Researchers aim to develop better brain implants for restoring vision by stimulating high-level parts of the visual system, potentially creating richer visual experiences for individuals with visual impairments.

🌏 China dominates global generative AI patent applications, with U.S. as distant second. Link

• China filed over 38,200 generative AI patent applications in the past decade, six times more than the U.S. with nearly 6,300.

• The World Intellectual Property Organization reports a surge in generative AI patents, with 54,000 inventions filed in the last decade and over 25% emerging in 2023 alone.

• While China leads in patent quantity, the U.S. outpaces in other metrics such as notable machine learning models (61 vs. China's 15 in 2023) and development of AI foundation models.

💻 Chinese buyers smuggle Nvidia chips to circumvent US export ban. Link

• Some Chinese buyers are using underground channels to smuggle advanced Nvidia GPUs into China, with travelers bringing them in their luggage to avoid US chip curbs.

• Distributors and brokers are facilitating the smuggling, with one Beijing-based distributor claiming to receive dozens of chips monthly through these methods.

• The demand for Nvidia's advanced chips in China remains high due to their crucial role in training AI models, despite the US ban on advanced Nvidia chip imports implemented in 2022 and tightened controls in 2023.

Have any feedback? Send us an email