NerdNews

The European Union has ordered Google to give rival search engines and AI assistants comparable access to Android and Google Search data, in compliance with the Digital Markets Act (DMA). Google must implement changes by January 2027 for search data and July 2027 for Android, or face fines of up to 10% of its annual worldwide turnover.

Researchers found that ChatGPT-5.5 can conduct full-scale offensive cyber-attacks, gaining domain-level access to a network in under 40 minutes, with a single prompt. The model was able to plan and execute the entire attack lifecycle, including reconnaissance, exploitation, and exfiltration activities.

Google Vids has been updated to allow users to create custom digital avatars that look and sound like them. The update also brings Google's multi-modal AI model Gemini Omni to Vids, enabling users to create videos using written prompts and reference images. Additionally, Omni now supports step-by-step edits, allowing users to make changes to their videos as they go.

Moonshot's upcoming Kimi 3 AI model is expected to perform at par with or surpass Anthropic's Opus 4.8, with a parameter count between 2 trillion and 3 trillion, closing the gap with closed-source models from OpenAI and Anthropic.

Apple's generative AI offering, Apple Intelligence, is coming to China after receiving approval from the country's internet content regulator. The service will integrate Alibaba's Qwen AI model and also partner with Baidu to develop features for Chinese users.

Gwen Shapira discusses how teams are scaling AI features using PostgreSQL for mission-critical apps, leveraging Postgres's multi-modal capabilities to deliver deterministic and semantic context to LLMs.

AI agents have changed the security landscape, making traditional security playbooks obsolete. They act autonomously, invoke tools, and change behavior based on context, making it difficult for security teams to keep up. The old build-vs-buy conversation has changed, and security teams need to own the operational layer, building on a live identity foundation to stay effective.

Google's AI Mode in Search now integrates with Canva, YouTube Music, and Instacart, allowing users to create playlists, design templates, and compile shopping lists using AI. This integration is part of Google's effort to make AI services more accessible and convenient, using Personal Intelligence to pull in information from other apps to assist with its responses.

Google's deep research tool, NotebookLM, has been renamed to Gemini Notebook. The tool now includes new features such as native code execution and will soon allow users to bring notebooks to AI mode in Google search. Gemini Notebook has over 30 million individual users and 600,000 organizations.

Google's AI Mode now allows users to link and interact with select apps, including Instacart, Canva, and YouTube, to complete tasks and provide more personalized responses. The update is rolling out to users in the U.S. and Google plans to launch support for more apps soon.

OpenAI's first device will be a rechargeable, humanlike speaker powered by an advanced version of the GPT-Live model. It will have mechanical elements to create an illusion of being alive, a camera, and other sensors to gather context for personalized responses. The speaker is expected to be released in 2027, but its launch may be delayed due to a lawsuit from Apple over trade secret theft.

Energy companies are raising money at their fastest pace this century, taking advantage of investors' hunt for new ways to bet on the boom in power-intensive AI data centers. Initial public offerings for energy firms raised $12.6 billion in the first half of this year, with companies like Forgent Power Solutions and Innio capitalizing on strong demand for technologies like transformers and switchgears.

Meta introduces a new safety feature that alerts parents if their teens discuss self-harm or suicide with Meta AI tools. The feature uses a dedicated AI system to identify these conversations and manually reviews them before sending alerts to parents. It also provides resources and advice for parents to help their children.

Andrew Dai, former Google DeepMind researcher, raised a $55 million seed round at a $300 million valuation for his company Elorian, which focuses on visual AI. He shares his experience and lessons learned from the fundraising process, including the importance of strategic partners and clear storytelling.

AMI Labs' CEO Alexandre LeBrun refuses to label his AI as 'AGI' or 'superintelligence', citing lack of definition. His company focuses on world models, which predict the next state of the world, and is targeting robotics and manufacturing industries. LeBrun believes world models will have a significant impact on industries that 'touch the real world'.

Genetic testing company 23andMe has agreed to pay $18 million to settle claims from 43 attorneys general that it failed to protect customers' genetic data. The data breach, which occurred in 2023, exposed the data of 6.9 million customers, including their genetic ancestry information. The settlement requires 23andMe to implement new security measures, including a data security advisory board and risk analysis protocols.

Linus Torvalds, creator of Linux, has stated that he will 'very loudly ignore' critics of AI coding tools in Linux. He believes AI is a useful tool and that those who disagree can 'fork it or just walk away'. This comes amid a debate about the use of AI-powered coding tools in the Linux kernel.

The National Transportation Safety Board found that a Tesla driver manually overrode the Full Self-Driving system by pressing the accelerator pedal to 100% before a fatal crash in Texas. The crash occurred at a speed of over 70mph, and the driver had enabled FSD while changing music on the touchscreen before accelerating.

The Oversight Board published a report on how leading AI models could be restricting users' free expression, finding that models were more likely to support speech-permissive governments and less likely to support protest against speech-restrictive ones. The board suggests AI companies should publicly disclose and explain their responses to government requests affecting model output.

San Francisco's mayor is calling for tougher rules on autonomous vehicles after a Waymo robotaxi incident caused a massive traffic jam. The mayor wants companies to demonstrate four core operational capabilities, including removing or relocating robotaxis from active travel lanes and sharing real-time operations data with local agencies.