Daily AI & Dev Digest: Apple's AI Choices, White House Regulation, and Image AI's Growth Surge
Stay updated with the latest in AI: Apple's iOS 27 offers diverse AI models, CopilotKit empowers app-native agents, the White House considers tighter AI regulation, image AI drives app growth, and Tether AI innovates on edge devices.
Welcome to your daily dive into the rapidly evolving world of AI and software development! Today's headlines bring a mix of exciting user-centric advancements, crucial regulatory considerations, and innovative infrastructure plays, highlighting the multifaceted growth of artificial intelligence.
TL;DR
- Apple's iOS 27 is set to empower users with a 'choose your own adventure' approach to on-device AI models, offering choices from third-party LLMs.
- CopilotKit successfully raised $27 million to advance app-native AI agents, moving beyond traditional chatbots for smoother user experiences.
- The White House is reportedly considering a new working group to implement tighter regulation and federal review for emerging AI models.
- Image AI models are now significantly driving mobile app growth, generating 6.5 times more downloads than traditional chatbot upgrades.
- Tether AI is building a 'Stable Intelligence layer' designed for efficient scaling on edge devices, aiming to democratize AI access.
Apple plans to make iOS 27 a Choose Your Own Adventure of AI models
Apple is reportedly planning to offer iPhone users a significant degree of choice regarding the artificial intelligence models integrated into iOS 27, slated for release later this year. A report from Bloomberg indicates that users will be able to select from various third-party large language models to power functions within the operating system.
This new capability, internally referred to as "Extensions," will allow users to access generative AI features from installed apps on demand through existing Apple Intelligence functionalities like Siri, Writing Tools, and Image Playground. The feature is also expected to be available for iPadOS 27 and macOS 27. Initial testing reportedly includes models from Google and Anthropic, signaling a more open approach to AI integration on Apple devices.
The new feature, dubbed “Extensions” internally, will allow users to “access generative AI capabilities from installed apps on demand, through Apple Intelligence features such as Siri, Writing Tools, Image Playground and more.”
CopilotKit raises $27M to help devs deploy app-native AI agents
CopilotKit, a company focused on enhancing AI agent deployment, has successfully raised $27 million to further its mission. The company aims to move beyond the limitations of simple chatbot UIs, which often provide clunky, text-heavy experiences within applications.
Co-founders Atai Barkai and Uli Barkai envision a future where AI agents are deeply embedded within applications, capable of understanding user actions, performing tasks, and presenting intuitive interfaces rather than lengthy text responses. Their widely adopted, open-source AG-UI protocol is a key component in realizing this vision, enabling a more seamless and effective integration of AI into user workflows.
The company’s popular AG-UI protocol is aimed at the first part of that solution. The widely adopted, open-source protocol standa.
The White House Is Considering Tighter Regulation Of New AI Models
The White House is reportedly exploring the implementation of stricter regulations for new artificial intelligence models. According to a report from The New York Times, discussions are underway regarding the creation of a new working group to oversee AI development.
One potential power for this committee includes a federal review of new AI models before their public release, possibly mirroring oversight models seen in the UK government. This consideration marks a potential departure from the previously hands-off approach outlined in the White House's AI Action Plan, which had been criticized for offering concessions to AI companies while potentially creating new problems. The move comes as the technology industry faces numerous lawsuits related to AI. While no definitive approach has been decided, the possibility of an oversight group suggests a growing recognition of the need for safety standards in AI development.
A federal review of new AI models ahead of their public release is being considered as a possible power for that committee, according to the publication's sources.
Image AI models now drive app growth, beating chatbot upgrades
New data from app intelligence provider Appfigures reveals a significant shift in what drives mobile app growth: image AI models are now outperforming traditional chatbot upgrades. The report indicates that releases of image models generate 6.5 times more downloads for AI mobile apps compared to updates focused on conversational experiences.
This marks a notable change from earlier trends, where new conversational models and features like voice chat interfaces were the primary drivers of demand. For instance, both ChatGPT and Google’s Gemini saw substantial increases in downloads after introducing their respective image models. Specifically, Google’s Gemini experienced an additional 22+ million downloads in the 28 days following the introduction of its Gemini 2.5 Flash image model last August, demonstrating the powerful impact of visual AI capabilities on user adoption.
Image model releases are driving growth for AI mobile apps, generating 6.5 times more downloads than traditional model updates.
Tether AI is building the Stable Intelligence layer, a highly efficient platform designed to scale on edge devices, made for the people
Tether AI is developing a novel platform called the "Stable Intelligence layer," aiming to provide highly efficient and scalable artificial intelligence on edge devices. This initiative seeks to address the challenges faced by companies, particularly small to midsize enterprises, in fine-tuning or retraining AI models due to high costs and reliance on centralized cloud infrastructure.
According to Paolo Ardoino, CEO of Tether, their QVAC SDK and Fabric tools are designed to empower individuals and companies to execute inference and fine-tune powerful models on their own hardware, giving them full control over their data. This approach aims to question the net throughput of AI by making the technology more cost-efficient and accessible, circumventing the saturated ecosystem of LLMs competing for GPU compute resources in data centers.
“QVAC SDK and Fabric give people and companies the ability to execute inference and fine-tune powerful models on their own terms, on their own hardware, with full control of their data.” Paolo Ardoino, CEO, Tether.