Tech • AI • Robotics
AI is playing with our lives and Apple, Meta and Google are going crazy
9.4/10TL;DR
Warnings from a former Anthropic and OpenAI researcher, major Apple hardware launches, Meta’s new AI agent, Chinese model gains, and a European shake-up of Google Search all highlight how fast artificial intelligence and digital platforms are reshaping markets and regulation.
KEY POINTS
AI safety alarm goes mainstream
Former researcher Jacob Coxon, who worked on pre-training at OpenAI and Anthropic, resigned while accusing leading AI labs of building systems they do not fully understand or control. His criticism focused on increasingly opaque training and self-improving AI loops, where models help supervise, refine or generate future model behavior. The warning spread widely online, but the core issue is less a sudden revelation than a fresh signal that capability gains are outpacing oversight.
Anthropic’s internal view adds weight
The debate intensified because current Anthropic alignment researcher Evan Hubinger publicly backed much of the concern. He said the company genuinely considers extreme AI catastrophe plausible and personally estimated the risk at more than 10% within the next decade. Anthropic has already published risk reports and research on recursive improvement, showing that these dangers are being discussed inside labs rather than ignored.
The real fault line is the race
The central issue is not one company alone but the global incentive structure. Labs are under pressure to release stronger systems quickly, while governments have yet to establish robust international rules on deployment, auditing and limits. Calls for pauses or regulation remain politically difficult because the firms asking for caution often also benefit from preserving their lead.
Apple unveils the iPhone Duo
Apple used its September launch to introduce the iPhone Duo, a foldable device positioned as both smartphone and tablet. It features a 7.6-inch inner display, a roughly 5.25-inch outer screen, under-display front camera technology and Touch ID moved to the edge. Pricing starts at $1,999, with preorders from 16 October and availability from 23 October.
Apple doubles down on on-device AI
The company also presented the iPhone 18 Pro and 18 Pro Max, built around the A20 Pro chip in 2 nanometers with a 6-core CPU and 7-core GPU. Apple said graphics performance rises by up to 20%, with some GPU tasks improving by up to 40%, while AI features are increasingly processed locally. The company is also expanding provenance tools to mark AI-generated or AI-edited media.
Health features expand, but Europe waits
New Apple Watch Series 12 and Ultra 4 models push further into health and coaching, with more precise heart monitoring, heart-rate variability metrics and tighter AI integration through watchOS 27. But several AI and health-related functions will arrive later in Europe. The delay reflects Apple’s caution around European digital and health regulation rather than an outright ban.
Meta turns AI into an action agent
After releasing a series of Muse models across chat, image, video and code, Meta is now packaging them into a consumer-facing assistant designed to do tasks rather than just answer questions. Muse is being positioned inside apps such as WhatsApp and mobile interfaces, with integrated browsing to handle bookings, forms, messages and listings. One flagship example is creating and posting a used-car sale from a simple photo.
Chinese models keep up pressure
Competition is also accelerating from China, where a newly released model was described as outperforming recent offerings from major US companies including Meta and Google only days after earlier launches. The broader trend is toward cheaper, lighter and more efficient systems, not just raw benchmark gains. That shift matters because lower cost can drive faster adoption than prestige “pro” models.
Google Search in Europe changes dramatically
A major update to classic Google Search in the 27 EU countries is reshaping visibility for hotels, travel firms and merchants. To comply with European rules, Google is reducing direct integrations of its own services such as rich maps, price modules and native recommendation panels. Google says the result is the biggest degradation in search quality in its 29-year history, arguing that users lose convenience while intermediaries gain prominence.
AI summaries may become the new gatekeepers
The unintended consequence is that classic organic traffic is squeezed from two sides. Traditional search modules send less direct traffic to merchants, while AI overviews answer more queries without a click. Yet those same AI summaries can still elevate direct providers, meaning the power to route attention is shifting from classic search ranking toward conversational AI systems including Gemini, ChatGPT and Claude.
CONCLUSION
Across AI labs, consumer devices and search platforms, the common theme is acceleration without stable rules. The next phase of competition may depend less on who builds the most powerful model than on who controls the interface, regulation and trust around it.
Explain this