NerdNews

Google Photos has introduced new AI-powered features, including object and people editing, AI templates, and a redesigned photo editor. The AI-powered search feature is also expanding to over 100 countries, supporting 17 new languages. The updates include a new Ask button for AI requests and personalized edits based on face groups.

The article discusses how AI is not displacing High Performance Computing (HPC) but rather is a subset of it. The author argues that AI and HPC have similar requirements and that the industry supporting HPC is not in danger. The real challenges lie in coping with the increasing power consumption and heat generation of AI systems, which can be addressed through efficient IT infrastructure and liquid cooling.

OpenAI is projected to lose $74 billion in 2028, while Anthropic will break even the same year. OpenAI has committed $1.4 trillion to building its data center infrastructure over the next 8 years and is expected to burn about 14 times as much money as Anthropic before turning a profit by 2030.

Top tech leaders like Elon Musk and Meta's Mark Zuckerberg predict superintelligent AI will arrive soon, but a new study finds that the average expert disagrees, giving the rapid progress timeline only a 23% chance of happening. Experts predict AI will have a significant impact by 2040, but not as soon as tech leaders claim.

Anthropic released Claude Haiku 4.5, a hybrid reasoning large language model with faster performance and lower costs. The model is available through multiple platforms, including Anthropic's API, Amazon Bedrock, and Google Cloud's Vertex AI. It features extended thinking mode, context awareness, and precise information about memory consumption.

Yann LeCun, Meta's chief AI scientist, plans to leave the company to build his own startup focused on world models, an AI system that develops an internal understanding of its environment. LeCun's departure comes as Meta overhauls its AI organization and invests in new technologies.

The European Commission's proposed reforms to the GDPR have been leaked, and privacy advocates are condemning the plans, accusing officials of favoring Big Tech interests. The reforms would introduce loopholes allowing companies to use personal data for commercial benefit, weaken protections for sensitive data, and make it easier for data controllers to comply with data protection laws while using people's data to train AI models.

Google introduces Private AI Compute, a secure cloud platform for advanced AI features while maintaining user data privacy. This platform allows for more complex AI processing, enabling features like personalized suggestions and expanded language support, without compromising on-device data security.

ElevenLabs launches an AI marketplace where companies can license AI-replicated voices of famous figures, including Michael Caine and Mark Twain, for use in ads and content, with a focus on consent and fair compensation

Steven Adler, who led product safety at OpenAI, expresses concerns about the company's decision to allow erotic content on its platforms, citing potential mental health risks and lack of transparency. He discusses the need for standardized safety benchmarks and more rigorous testing of AI systems.

Kaltura, a video platform company, has acquired eSelf, a startup behind conversational avatars, for $27 million. The acquisition will enable Kaltura to integrate eSelf's virtual agent technology across its video offerings, allowing for real-time, synchronous conversations. eSelf's technology supports over 30 languages and features a user-friendly studio for creating and deploying photorealistic digital avatars.

Wonderful, an AI agent startup, has raised $100M in a Series A round to deploy customer-facing AI agents across voice, chat, and email in every market and language, with plans to expand into new areas like employee training and sales enablement.

Robyn is an AI companion designed to provide emotional support and understanding. It is not meant to replace human relationships or therapy, but rather to offer a supportive and non-judgmental space for users to express themselves. The app uses a unique emotional memory system to understand and respond to users' emotions, and it has been designed with safety features to prevent users from anthropomorphizing the chatbot.

Broadcom and CAMB.AI have developed an AI chipset that enables on-device audio translation in over 150 languages, reducing latency and enhancing privacy. The chipset can also provide audio descriptions of on-screen content, making it useful for individuals with vision issues.

Nokia Bell Labs and KDDI have entered a joint research agreement to explore new approaches for making 6G networks more energy efficient and resilient. The research will focus on two key areas: mobile multiple input, multiple output (mMIMO) energy efficiency and distributed programmable core network services for 6G.

The AI bubble is a complex issue due to mismatched timelines between AI software development and data center construction. Companies like Oracle and Meta are investing heavily in AI infrastructure, but there's uncertainty about demand growth and potential infrastructure problems. A McKinsey survey found that most companies are still in 'wait and see' mode with AI adoption.

Google's November Pixel Drop update brings new features including a power-saving mode for Maps, AI-powered photo editing, and smarter notifications. The update also expands support for call transcription and introduces a 'Remix' feature for reimagining photos using prompts.

Meta's top AI scientist, Yann LeCun, is reportedly leaving the company to launch his own startup. LeCun, a Turing Award winner, is a leading figure in modern AI and was a key part of Meta's AI team. His departure is the latest blow to Meta's AI ambitions, which have struggled in recent months.

North Korean state-backed spies have exploited Google's Find My Device feature to remotely wipe Android phones, destroying evidence and disrupting targets. The KONNI group used stolen Google account credentials to access victims' profiles and trigger factory resets, highlighting the risk of 'lost device' features tied to online identity systems.

Nokia Bell Labs and KDDI have entered a joint research agreement to explore new approaches for making 6G networks more energy efficient and resilient. The research will focus on two key areas: mobile multiple input, multiple output (mMIMO) energy efficiency and distributed programmable core network services for 6G.