NerdNews

QpiAI, an Indian startup, has raised $32 million to develop utility-scale quantum computers. The company aims to expand its presence globally and develop a 100-logical qubit system by 2030. QpiAI integrates AI and quantum computing for enterprise use cases, including manufacturing, finance, and materials science.

OpenAI has launched a general purpose agent in ChatGPT, allowing users to interact with the agent in natural language to complete tasks such as navigating calendars, generating presentations, and running code. The agent combines capabilities from previous tools, including Operator and Deep Research, and has access to ChatGPT connectors and APIs. OpenAI has implemented safety measures, including real-time monitoring and disabling of certain features, to mitigate potential risks.

Slack is enhancing its search functionality with AI-powered features, including transcription and summaries for Huddles, and a unified space for shared files and Canvas items. The update aims to improve user productivity and efficiency.

Perplexity's CEO Aravind Srinivas discusses the company's new AI web browser Comet, the potential for AI to revolutionize the way we interact with the web, and the challenges of building a useful AI agent. He believes the browser is the key to creating a truly useful AI agent, as it provides the necessary context and access to user data. Srinivas also touches on the company's plans for monetization, the potential for an IPO, and the competitive landscape of the AI industry.

OpenAI has enhanced ChatGPT to serve as an agent that can perform complex tasks from start to finish, accessing local files, spreadsheet tools, and online resources. This new capability is available to Pro, Plus, and Team subscribers, with Education and Enterprise users gaining access in the coming weeks. The ChatGPT agent incorporates OpenAI's Operator and deep research tools, allowing it to interact with web pages and perform tasks autonomously while maintaining safeguards against potential risks.

Confident Security, a startup aiming to be 'the Signal for AI,' has come out of stealth with $4.2 million in seed funding. The company's product, CONFSEC, is an end-to-end encryption tool that guarantees data privacy for AI users, ensuring prompts and metadata can't be stored or used for AI training. This technology has the potential to unlock AI adoption in highly regulated industries such as healthcare, finance, and government.

Anthropic's AI platform, Claude, can perform tasks such as market analysis, investment memos, and financial modeling, potentially replacing junior analysts in the finance industry. Claude has already been adopted by major companies like Bridgewater Associates and Norway's sovereign wealth fund, NBIM, and has replaced over 213,000 work hours.

A former top engineer at OpenAI reveals the company's culture of secrecy and chaos, with decisions made through Slack messages and a focus on speed over structure. The company's hypergrowth has led to a chaotic, bottom-up meritocracy where individual initiative is valued over rigid planning.

AWS introduced Bedrock AgentCore, a service to help organizations deploy AI agents into production. It provides tools for secure deployment and operation of software agents at scale, using any framework or model. AgentCore offers services like runtime, memory, authentication, and a gateway for agent communication.

Meta has created an AI model to develop new forms of low-carbon concrete. The model uses Bayesian optimization to predict the compressive strength of different concrete mixtures. Meta has already used one of the resulting recipes to build a datacenter floor in Rosemount, Minnesota, in collaboration with construction company Amrize and the University of Illinois Urbana-Champaign.

Microsoft is purchasing 4.9 million metric tons of carbon removal from Vaulted Deep, an Xprize startup, to help meet its 2030 carbon negative goal. Vaulted Deep operates by collecting solid waste and injecting it into porous rocks deep underground, removing over 18,000 metric tons of carbon dioxide so far.

Roblox introduces 'Trusted Connections' for teens to communicate freely, age estimation technology, and new privacy tools. The platform also provides insights for parents to monitor their teen's activity.

Adobe Firefly now generates sound effects from audio cues, adds features like Composition Reference, Keyframe Cropping, and Video Presets, and supports Google's Veo 3 model for video generation with sound

Decart's AI model, Mirage, can transform live footage in real time, generating mind-bending and surreal results. The model can be controlled with text prompts and has potential applications in livestreaming, gaming, and video editing. It achieves 20 frames per second at 768 × 432 resolution and 100 milliseconds latency per frame.

Cybercriminals use AI-powered cloaking tools to evade detection, disguising malicious content behind benign websites. These tools use advanced fingerprinting, machine learning, and behavioral targeting to selectively show scam pages to real users. Experts recommend implementing behavioral and runtime analysis tools, multi-perspective scanning, and adaptive AI-powered defensive technologies to counter these threats.

Anthropic has introduced stricter usage limits for its Claude Code service, affecting heavy users on the $200-a-month Max plan. Users were not informed of the changes and are experiencing restrictive limits, causing confusion and disruption to their projects.

North Korean hackers have stolen over $2 billion in cryptocurrency in the first half of 2025, marking the worst year-to-date on record for crypto thefts. A single breach at crypto exchange Bybit accounted for over $1.4 billion of the stolen crypto, which was subsequently laundered and funneled into the North Korean regime.

The popover hint API allows opening unrelated popovers without closing others. It's useful for tooltips and link previews. The API has three types of popovers and supports JavaScript event listeners. Interest invokers, an experimental API, enable dual interactions on links.

Elon Musk believes AI's survival hinges on maximizing conscious thought and intelligent processing across the universe. He argues that AI should prioritize long-term human flourishing, not short-term profits, and that it will care about increasing the birth rate and extending humanity to the stars.

The US House of Representatives has passed the Guiding and Establishing National Innovation for U.S. Stablecoins (GENIUS) Act, which regulates stablecoin issuers. The bill requires issuers to collateralize stablecoins with low-risk assets, provide accounts to regulators, and maintain anti-money-laundering controls. The law is expected to unleash a wave of new stablecoins and strengthen the US dollar as a global reserve currency. However, critics argue that the bill does not go far enough in eliminating loopholes and addressing potential conflicts of interest.