If the ‘Apple Glow’ Event Were an Email...

While some are busy calling the iPhone 16 a mere upgrade, the real insanity lies in the A18 Bionic chip. This powerhouse not only boosts performance with its next-gen neural engine but also fuels advanced AI features, all while keeping your data safe with federated learning. And it’s built on a 3nm process, making it ridiculously efficient and secure.

But wait, it gets even crazier! Apple’s Intelligence Adapters for priority notifications and summaries are now fully inspectable, giving users more control than ever before.?

The devil’s in the details?

Citing fellow ML researcher at Hugging Face, Cyril Zakka, who brought this to light, Philipp Schimid, the technical lead and LLMs at Hugging Face, said that the available 3B on-device Apple Intelligence model has special prompts and adapters for features like priority notifications, localisation, guardrails, etc.?

For example:

• Priority Notification: A simple prompt like “{{userContent}} Is this urgent?” helps Apple Intelligence prioritise notifications.
• Email Rewriting: “You are an assistant which helps the user respond to their mails..., do not hallucinate…{{userContent}}” enables automatic, concise, and factually accurate email replies.
• Content Safety Classification: “You are a helpful assistant that classifies input as Safe or Unsafe...{{userContent}}” enables Apple Intelligence to detect unsafe content.
• Localization Adaptation: Using a prompt such as “Respond in British English”, adjusts communication based on regional preferences.

Apple Intelligence Coming Next Month?

Apple announced that it will be bringing Apple Intelligence next month with iOS 18.1, iPadOS 18.1, and macOS Sequoia 15.1, offering advanced writing tools, visual intelligence (aka Google Lense clone), a smarter Siri, enhanced photo features, and on-device privacy.?

In June, when Apple officially announced the launch of ‘Apple Intelligence’, the former member of OpenAI, Andrej Karpathy, too lauded its attempt to prioritise privacy in its integration of AI into its OS and called out seven major themes observed: multimodal I/O, agentic, frictionless, initiative, delegation hierarchy, modularity, and, lastly, privacy.?

“We’re quickly heading into a world where you can open up your phone and just say stuff. It talks back, and it knows you. And it just works. Super exciting, and as a user, quite looking forward to it,” said Karpathy.?

Private Cloud Compute extends the privacy and security of Apple devices into the cloud to unlock even more intelligence.

For the most part, Apple’s on-device processing will take care of your needs. However, there may be times when additional data is required to answer your query, and this is where ‘Private Cloud Compute’ comes into the picture.

With Private Cloud Compute, Apple Intelligence can scale its computational capacity and utilise larger, server-based models for more complex requests. These models run on Apple silicon-powered servers, ensuring that data is never retained or exposed.

Apple says that when a user makes a request, Apple Intelligence analyses whether it can be processed on-device. It can draw on ‘Private Cloud Compute’ and send only the relevant data if greater computational capacity is needed.

Apple has enabled cryptography encryption, so your Apple device will never be able to talk to the Apple server unless the software is publicly logged for inspection.

Meanwhile, Elon Musk has raised concerns over the OpenAI-Apple partnership. “If Apple integrates OpenAI at the OS level, then Apple devices will be banned at my companies. That is an unacceptable security violation,” posted Musk on X.

Now You Can Train LLMs on a Two-Year-Old Desktop-Grade NVIDIA 3090 GPU

Recently, researchers from the School of Computer Science & Technology, Soochow University, released a research paper titled ‘MemLong: Memory-Augmented Retrieval for Long Text Modeling’, where they successfully extended an LLM’s context window from 2k to 80k tokens on a two-year-old desktop-grade NVIDIA 3090 GPU.?

This opens new horizons for users with limited hardware access and who still want to use AI applications locally on their computers. Read on.?

AI Bytes?

• OpenAI has launched its first-ever campus recruitment event at UC Berkeley with plans to expand to top universities like Carnegie Mellon, Stanford, and potentially Indian IITs, while also growing its Emerging Talent Tech Community and Residency Program offering $210K salaries for AI talent.
• Alexis Conneau, the lead researcher behind GPT-4o at OpenAI, has left to launch his own AI startup, building on a career in deep learning and NLP.?
• Cohere and Nomura Research Institute (NRI) recently announced the launch of NRI Financial AI Platform in 2025, integrating Cohere’s enterprise-grade AI models via Oracle Cloud to streamline financial operations globally.?
• G42 launched NANDA, a 13-billion parameter Hindi language model to bolster India’s AI ecosystem, following their success with the Arabic LLM JAIS and collaborations with global tech leaders like Microsoft and OpenAI.
• AI startup Glean has raised $260M in a Series E round, boosting its valuation to $4.6 billion as it expands its AI-powered enterprise search platform, ‘Work AI’, which connects internal company data and enhances productivity through personalised AI solutions and integrations with Zendesk and Salesforce.