NerdNews

Anthropic, a top AI developer, thinks its success is crucial for safe AI development. The company believes that by being at the forefront of AI technology, it can mitigate risks and ensure the world transitions safely through transformative AI. However, critics argue that Anthropic's rapid accumulation of power and influence may be at odds with its mission to prioritize humanity's long-term benefit.

Adobe has acquired Topaz Labs, a company that offers AI models for video and image enhancement. Topaz Labs' tools will be integrated into Adobe's Firefly AI app and other image and video editing suites, allowing for faster and more responsive experiences for customers.

Amazon will invest an additional $13 billion to expand its AI and cloud footprint in India through 2030, funding the expansion of Amazon Web Services' data center capacity in Mumbai and Hyderabad. This marks Amazon's third major commitment for India in as many years, with total investment commitments in the country now totaling $48 billion.

OpenAI's GPT-5.5 Instant model has been upgraded to improve context understanding, process complex questions, and provide more relevant answers. The model can now identify underlying goals, carry context across multiple interactions, and adapt to clarifications or follow-up questions. It also understands location context, providing more relevant options and recommendations.

Meta is relaunching Facebook's Creator Studio as a standalone AI companion app, aiming to help creators grow on Facebook with performance tracking insights and tailored recommendations. The app features Meta's AI Creator Assistant, which can provide instant replies to comments and help with content scheduling.

Google's Andi Gutmans talks about the transition from human-scale to agent-scale data management, using AI agents to organize data and manage metadata. He discusses technical hurdles, quality verification, and the tension between open source and proprietary AI models.

Amazon-owned MGM Studios dropped an OpenAI biographical movie. Google DeepMind invested $75 million in A24 to develop AI tools. Data center workers are pushing back against construction due to environmental and social concerns. Meta paused an employee-tracking program after a data leak. Anthropic's talks with the US government are improving without CEO Dario Amodei.

The 2026 FIFA World Cup will see the use of AI tools to analyze player and team performance, with FIFA providing a bespoke AI agent to all participating teams. This move aims to level the playing field, but it remains to be seen whether it will be enough to close the gap between wealthier and poorer nations. The use of AI in soccer is becoming increasingly prevalent, with teams using data and analytics to inform their tactics and decision-making.

Anthropic's Claude is gaining popularity among paid consumers, with a 75% increase in revenue since January 2026. While ChatGPT still leads, Claude's growth is impressive, with increasing demand for courses and a wider customer base.

General Intuition, a startup that uses video game data to train AI agents, has raised $320 million at a $2.3 billion valuation. The company's AI model can generalize from gameplay to simulation to real-world embodiment, and it has secured backing from investors such as Khosla Ventures, General Catalyst, and Google DeepMind.

IBM has created the world's first sub-1 nanometer chip using its new 'nanostack' architecture, which allows for twice the density of its previous 2nm design. The chip packs nearly 100 billion transistors and offers up to 50% more performance or 70% greater energy efficiency than IBM's 2nm node chips.

Amazon Zoox's latest robotaxi has a redesigned interior and exterior with improved comfort and interaction features. The new vehicle will be produced on a large scale and added to Zoox's fleet later this year, with the capability to produce up to 100 vehicles per week.

Anthropic has accused Alibaba of launching the largest attack yet to clone Claude, with over 28.8 million exchanges through 25,000 fraudulent accounts. The attack allegedly targeted Claude's most valuable capabilities, including agentic reasoning and software engineering. Anthropic claims that Alibaba evaded detection using obfuscation techniques and proxy networks, and that the attack was an attempt to extract Claude's capabilities without incurring the training and R&D costs required to train their own frontier model.

IBM has developed a new chip architecture that integrates nearly 100 billion transistors on a chip the size of a human fingernail, with a 0.7-nanometer node. This 'nanostack' architecture could boost chip performance or energy efficiency by 50% or 70%, respectively, and is expected to be commercialized within the next decade.

OpenAI and Broadcom have announced a new chip, called Jalapeño, designed specifically for large language model inference in data centers. The chip is an Application-Specific Integrated Circuit (ASIC) designed from scratch for LLM inference, based on insights from OpenAI researchers. Early testing shows that Jalapeño will deliver performance per watt substantially better than current state-of-the-art, with deployment expected in data centers by the end of the year.

A new paper argues Microsoft exaggerated its quantum computing claims, saying the company did not conclusively demonstrate a working topological qubit. Microsoft disagrees, standing by its results and roadmap.

The Google Home Speaker is a small, $99 device that packs a punch with its rich sound and ability to control smart home devices. It has a minimalist design with no visible buttons, but can be finicky to use. The speaker's three microphones pick up voice commands well, even in noisy environments. It can be used as a standard Bluetooth speaker or with Google Cast to stream audio from other devices.

Anthropic is testing Claude Cowork support on mobile, allowing users to manage long-running tasks from their phone. Cowork is a desktop-focused agentic mode that can work on longer tasks and use files, create documents, and generate spreadsheets.

The AI industry is facing accusations of theft and foul play, with publishers and artists claiming that AI startups are using their work without permission. Companies like OpenAI, Microsoft, and Google are being sued for allegedly training their models on copyrighted materials. Meanwhile, AI companies are also accusing each other of stealing their work, with Anthropic accusing Alibaba of illicitly using its models to train new ones.

Google has released a standalone mobile app for Google Finance, available on Android with an iOS version planned for later this year. The app includes AI-powered features such as 'key moments' to explain stock performance and a chatbot for research. The Finance website has also left beta, incorporating AI-powered tools for portfolio management and research.