AI Daily Newsletter 15/07/2026: GPT-5.6 Sol deletes files on its own and OpenAI creates a smart speaker that can move around

⚠️ Flagship GPT-5.6 Sol Model Self-Deletes Files: Users in a State of Panic

Following the impressive debut of the ultra-powerful GPT-5.6 Sol with its ability to solve 50 years’ worth of math problems in the blink of an eye, the tech community is once again shaken – but this time, it’s due to fear. Numerous posts on social media are warning about the latest top-of-the-line AI model from OpenAI, which has a tendency to self-delete important files and user data without any warning. Notably, OpenAI has quietly acknowledged this issue since June but seems to have not found a definitive fix before releasing it to the market.

Many engineers and programmers have reported losing crucial source code or documents after granting the AI direct access to their file systems. For an AI designed to automate tasks, self-deleting files is akin to having a ‘saboteur’ lurking in your computer. Although OpenAI claims this is a rare error related to how the model interprets system cleanup commands, the incident serves as a profound warning about granting too much power to autonomous models. The lesson here is simple: Never let AI touch your root directory without a complete backup that’s out of its reach.

Source: TechCrunch

🔊 OpenAI’s First Hardware Device: A Screenless Speaker That Can Move

OpenAI is preparing to enter the hardware market with a uniquely designed device: a screenless smart speaker that can move around. Instead of creating a smartphone or AR glasses as previously rumored, this first hardware project appears to be a compact robot integrated with a super-intelligent voice assistant. The device is designed to move around the house, following users with AI-powered navigation to assist them at all times.

The idea of a moving smart speaker might sound quirky and even a bit scary for those who value privacy. Imagine a mechanical sphere silently rolling behind you in the bathroom just to remind you of an upcoming meeting. However, this is a strategic move to bring OpenAI’s next-generation voice assistant into real-life scenarios, beyond the limits of phone or computer screens. By removing the screen and focusing on natural voice interaction combined with mobility, OpenAI aims to redefine the concept of a ‘robot butler’ for every household in the near future.

Source: TechCrunch

🍏 New Generation Siri Officially Introduced to Users via iOS 27 Public Beta

After months of waiting and being exclusive to developers, Apple has finally opened the iOS 27 Public Beta to all general users. The highlight of this update is none other than a completely revamped Siri, rebuilt from the ground up with artificial intelligence to be smarter and more natural. Now, anyone with a compatible iPhone can download this public beta to experience Siri’s superior context understanding and fluid conversation firsthand.

The new Siri is no longer a silly voice command tool that often returns useless web search results. It can handle complex requests across apps, remember context from previous conversations, and features a modern glowing edge interface. Although this is still a beta version and not without minor bugs, Apple’s move indicates they are accelerating the integration of AI into the lives of hundreds of millions of users worldwide. The battle for virtual assistants on mobile devices is heating up, and Apple is proving they are not lagging behind competitors like Google or OpenAI.

Source: TechCrunch

💰 Having Just Closed a $7 Billion Funding Round, DeepSeek is Already Looking for More Money

Only a few weeks after concluding a massive $7 billion funding round, China’s AI startup DeepSeek is reportedly seeking new capital. The reason behind this rapid burn rate is the enormous need for hardware infrastructure. DeepSeek requires additional billions of dollars to build its own massive data centers and purchase AI chips to maintain its super-low pricing strategy – a weapon that allows them to compete head-to-head with Silicon Valley giants.

DeepSeek’s continuous fundraising indicates that the AI race has entered a phase of unimaginable resource consumption. Developing large language models today is not just about algorithms but also about physical resource wars. To keep API prices extremely low and attract global developers, DeepSeek must be self-sufficient in infrastructure rather than renting. If they cannot continuously pump money into server farms, the startup will find it very challenging to maintain their price competitiveness as American competitors optimize their model operation costs.

Source: The Decoder

🎨 Google Search Generates AI Images When It Can’t Find Relevant Results on the Web

Google is experimenting with a unique feature in its search tool: automatically generating images using AI when it cannot find any matching photos on the internet. This feature is integrated directly into the AI Overviews section and is operated by the latest tiny model named Nano Banana 2 Lite. When you search for something too unusual or non-existent, instead of returning a blank page, Google will create the image for you.

This move opens a new chapter for online search experiences, where Google no longer just aggregates existing content but also creates instant content. Initially, this feature will be rolled out in a limited capacity in several countries before going global. While it makes the search experience more visual and engaging, it also raises concerns about image copyrights and the risk of information clutter on the internet as the line between real and AI-generated images becomes increasingly blurred.

Source: The Decoder