IoT Chronicles: August 2025

Artificial intelligence is becoming centralised, peripheral devices are strengthening their security, and homes are becoming energy efficient - August saw some concrete changes. Microsoft is integrating GitHub into CoreAI, and Swissbit is launching ready-to-use, hardware-based security for older devices. Interested in more news? Read on.

‍

1. GitHub’s CoreAI Crossover: No New CEO, Tighter Microsoft Integration

GitHub is moving directly under the wing of Microsoft's CoreAI. Chief Executive Thomas Dohmke is stepping down to return to working with start-ups, but will remain with the company until the end of 2025 to help with the transition. Microsoft is not appointing a new CEO for GitHub instead the company's management will report directly to CoreAI, led by Jay Parikh. This change effectively ends GitHub's semi-independent structure since its acquisition in 2018 and tightens its alignment with Microsoft's central AI strategy.

📍The Verge

2. Plug-and-Play IoT Defence: Swissbit Reinvents Protection at the Hardware Layer

Hardware first IoT security with plug-and-play upgrades Swissbit argues that software-only security does not address the biggest risk, which is physical access. Therefore, it offers hardware security that can be implemented in existing projects, including microSD cards with AES 256 encryption and secure boot options. This will help protect edge data in new and older systems by preventing data tampering, data theft and malware injection on devices using SD or USB media. At the same time, it will help customers meet the requirements of the EU's Cyber Resilience Act and NIS2 without having to redesign products manufactured in Germany for the supply chain trust and iShield Key 2.  Providing FIDO-based authentication and physical access control in a single compact key, an interview with Roland Marx, Senior Product Manager at Swissbit, was published on 6 August 2025.

📍vmblog

3. GPT-5’s Built-In Thinking: From Chat to Software-on-Demand

Meet ChatGPT-5, now positioned as the flagship model and automatically selected as the default interface for users. It actively manages its own behaviour in conversation: choosing when a quick reply will do and when a more considered response is warranted. Beyond that, it promises sharper writing, stronger comprehension and better coding than previous releases, while staying more firmly grounded in facts to reduce unsupported answers. Health use cases get special attention: ChatGPT-5 can draw on its knowledge to help patients prepare for appointments. Voice interactions also feel more natural, adapting to the user’s speaking pace and style. Pricing stays accessible: 1 million input tokens cost $1.25, while 1 million output tokens cost $10.

For more details and curiosities, check the launch video.

📍GPT - 5

4. Matter 1.4: Set Up Once, Control Everywhere with Energy-Smart Automation

Matter 1.4 moves the smart home from a compatibility demo to a real platform. Enhanced Multi-Admin finally lets you set up once and control across Apple Home, Google Home and Alexa. An IP backbone over Wi-Fi, Ethernet and Thread plus upcoming Matter-certified routers boosts reliability, while security adds vendor ID checks, access restriction lists and certificate revocation. The headline upgrade is energy management: solar, batteries, heat pumps and water heaters coordinate to shift loads and could trim bills by about 20–30%. New categories like appliances and leak/freeze/rain sensors with shutoff valves expand automation, and command batching makes scenes run in sync. Adoption is still uneven, but Amazon plans Echo/Eero updates early next year and Google is improving Multi-Admin.

📍Wonder How To

5. Perplexity bets $34.5 billion on Chrome

How serious is a $34.5bn shot at buying Google Chrome from a three-year-old AI start-up? Perplexity AI, led by ex-Google/OpenAI alum Aravind Srinivas and backed by Jeff Bezos and Nvidia, has made an unsolicited offer with no funding details, right as a US judge may rule on whether Google must break up its search business. Google calls any Chrome spin-off “unprecedented” and harmful to consumers, while investors quoted in the piece label the bid a stunt and far below Chrome’s true value. Some suggest it could be worth up to 10× more.

📍BBC

6. From Prompt to Picture: Meet Gemini 2.5 Flash Image

‍

Google has introduced Gemini 2.5 Flash Image, a native model for image generation and editing that emphasizes speed, control and consistent results. It supports multi-image composition, precise natural-language edits and draws on Gemini’s world knowledge for more accurate outputs.

Developers can start using it now via the Gemini API and Google AI Studio, with enterprise access on Vertex AI. The release is in preview, moving to stable soon, and includes template apps to help you prototype quickly. All generated or edited images carry SynthID invisible watermarking for provenance. For hands-on examples and how to get started, the full article has the details.

📍Google for Developers

August saw numerous changes in the virtual world that directly affected its users. Generating and editing images with a single prompt, improvements in smart home management, and a new edition of one of the most popular AI tools are just some of the changes and inspirations that emerged this month. There are four months left until the end of the year. What changes await us? We will find out soon. :)

‍