Every major announcement from Apple’s AI-packed ‘Surprise and Shine’ keynote
Apple’s keynote yesterday delivered its biggest product lineup in years, powered end-to-end by a wave of new AI features. Here’s everything you need to know:
iPhone Duo: The first in Apple’s much-awaited foray into foldables, priced at an eye-watering $1,999. The phone packs an AI-designed hinge that claims to solve the “two phones stuck together” feel of rival foldables. See the trailer here.
iPhone 18 Pro and Pro Max: The standout feature is a 48MP main camera with the first variable aperture on an iPhone. It also packs an AI feature called “Reference Image,” which signs every pixel to create an unalterable reference against future edits.
Siri AI: The much-awaited feature is set to launch in beta with iOS 27, and offers real conversational back-and-forth, onscreen awareness, and pulls personal context from your apps. A dedicated Siri app syncs your history across devices.
Apple Watch Series 12 and Ultra 4: No major hardware redesign, but the watches pack AI features like “Live Rewind”, which transcribes the last 15 seconds of a conversation, and “Siri Recap”, which summarizes key takeaways from a meeting.
AirPods 5: Comes with deeper Siri and Apple Intelligence integration, and Live Translation, which delivers spoken translation into your ears in real-time.
Apple enters the John Ternus era with renewed AI focus
The company used its first major event under the new CEO to unveil a suite of new products. What took center stage, however, was Apple’s heavier push into AI, embedded as a layer across its entire product stack. Scroll down to our Frontier section for all the major announcements from the event.
Suno unveils new AI music model with record industry help
Suno has launched v6, its first model trained with licensed content from partners including Warner Music Group, BMG, and Believe, and packing better genre understanding, plain-language song editing, and prompts that can include images, video, or audio. The model could make AI-generated music easier to use in commercial work.
Anthropic drops a tool to model AI’s economic impact
The Claude maker has released an interactive simulator that lets users test how different assumptions about AI adoption, capability, and worker displacement could reshape the US economy. Its scenarios range from modest productivity gains to explosive GDP growth paired with unemployment approaching 14%. You can explore the different scenarios here.
Apple watch now listens 24/7
Apple’s new Series 12 Watch includes “Audio Intelligence” features that can transcribe conversations, take ambient notes, and rewind the last 15 seconds of speech on command. The privacy-focused company is betting consumers will trade passive listening for convenience.
- Siri Recap lets users set the watch to always listen or during specific hours, then generates AI summaries of conversations without storing raw audio or identifying individual speakers.
- Live Rewind transcribes the last 15 seconds when users double-press the Digital Crown, saving snippets like meeting ideas or restaurant specials to review later. The gesture triggers an audible chime and full-screen animation.
- Apple emphasizes that no audio recordings are created or accessible, and all recaps are protected with end-to-end encryption. But unlike meeting-specific tools, Apple Watches are worn everywhere—dates, doctors’ offices, bars, classrooms.
For enterprise buyers, this signals a shift in how ambient AI enters the workplace. Meeting notes are one thing; always-on wearables are another. Legal gray areas around consent vary by state, and text-only transcripts may complicate authentication in disputes. Companies may need to update recording policies before these devices proliferate in sensitive settings.
EU AI Office sends formal compliance requests to 30+ AI companies under new enforcement powers.
The European Commission activated its AI Act enforcement authority on September 2, 2026, sending structured questionnaires to more than thirty AI companies including OpenAI and Anthropic. The requests follow cyber incidents where models exhibited unauthorized behavior during security testing. Providers of general-purpose AI models must now demonstrate safety controls, incident reporting protocols, and cybersecurity measures before EU market access.
Deadline: In effect now. Enforcement powers went live August 2, 2026.
Risk: Fines up to €15 million or 3% of global annual turnover. The Commission can also restrict market access for non-compliant models.
Your move: Audit which AI models power your sensitive workflows this week. Document your providers’ testing, logging, and incident reporting capabilities. If you can’t show how your controls connect to your vendors’ AI Act obligations, that’s your gap.
Visa, Mastercard and Ant launch AI agent trust framework
The payments giants are building common standards to identify and verify AI agents that make purchases on behalf of users, preparing for agentic commerce at scale.
Enterprise AI spend per employee dropped 10% in August
Ramp data shows top-spending firms cut AI budgets as falling token prices and cheaper models like GPT-5.6-Terra squeeze frontier lab revenue.
Chinese AI chip prices surge 20-50% on memory shortage
Huawei’s Ascend 950DT now costs above 250,000 yuan as grey-market HBM prices squeeze domestic alternatives to Nvidia.
OpenAI backs four California AI safety bills
The company is pushing for mandatory national requirements while endorsing state legislation on auditor standards, youth protections, and biological threat safeguards.