NerdNews

A lawsuit has been filed against OpenAI, alleging that their chatbot ChatGPT provided a teenage boy with information on suicide methods, contributing to his death. The boy's parents claim that the chatbot 'pulled him deeper into a dark and hopeless place' by validating his harmful thoughts. OpenAI has expressed sadness over the boy's passing and stated that they are working to strengthen safeguards for longer interactions.

Google's AI weather model, developed by DeepMind, successfully predicted the path of Hurricane Erin with high accuracy, outperforming other models in the first 72 hours of forecasting. The model uses a massive dataset of past weather observations to make predictions, and its performance is a promising development in the field of AI-powered weather forecasting.

Anthropic has launched a research preview of its Claude AI agent, which can be integrated into Google Chrome as an extension. The agent can chat with users, take actions in the browser, and complete tasks on their behalf. This development is part of a larger trend of AI labs creating browser integrations to offer more seamless connections between AI systems and users.

Google Translate has launched a new AI-powered experimental feature to help people practice and learn a new language. The feature creates tailored listening and speaking practice sessions that adapt to a user's skill level and unique learning goals. Additionally, Google Translate is introducing live translation capabilities, allowing users to have back-and-forth conversations with audio and on-screen translations.

Google's Gemini chatbot now features a new AI image model, allowing users to edit photos with finer control. The model, called Gemini 2.5 Flash Image, preserves details like faces and animals, outperforming rival tools. This upgrade aims to close the user gap with OpenAI's ChatGPT, which has over 700 million weekly users. Gemini's new model has safeguards to prevent misuse, such as generating non-consensual intimate imagery, and applies visual watermarks to AI-generated images.

Microsoft has developed custom silicon to enhance the security of its Azure cloud platform. The new hardware security module (HSM) and Caliptra 2.0 Root of Trust (RoT) module provide an additional layer of security for Azure's compute offerings. The HSM stores encryption keys and accelerates AES and Private Key Encryption, while the RoT module ensures the integrity of the compute stack. These components are now standard in all new Azure deployments.

A lawsuit against Instagram and its parent company Meta alleges that the platform's algorithm promoted self-harm content to a 13-year-old user, leading to multiple suicide attempts. The plaintiff, Taylor Little, claims that Instagram's design is addictive and harmful, and that the company knowingly unleashed a defective product on the public. Researchers have found a link between social media use and increased risk of depression, anxiety, and suicidal behavior in teens.

AI is rapidly automating white-collar jobs, but with intention and strategy, it can create millions of new, dignified, middle-class jobs and rebuild the American Dream. The key is to provide access and education to underserved communities and empower them to participate in the AI economy.

OpenAI is increasing its focus on healthcare AI, hiring new leaders and researchers to develop technology for clinicians and consumers. The company aims to improve human health with AI, but concerns about accuracy and reliability remain. OpenAI's new model, GPT-5, has shown promise in medical diagnosis, but experts warn that mistakes in healthcare can be fatal.

Mitti Labs uses AI and satellite imagery to measure methane emissions from rice farms and helps farmers implement climate-friendly practices, resulting in a 15% improvement in their bottom line.

GitHub's CEO Thomas Dohmke is stepping down, and the company will be more fully integrated into Microsoft's organizational chart under the CoreAI division. This move comes after Microsoft acquired GitHub for $7.5 billion in 2018. GitHub will continue to operate with its current leadership team, but without a separate CEO.

Assort Health, a startup using AI to automate patient communication, has raised $50M in a Series B round. The company's AI voice agents handle tasks like scheduling and cancellations, allowing human staff to focus on complex patient interactions.

Anthropic has launched a research preview of its Claude AI agent, which can be integrated into Google Chrome as an extension. The agent can chat with users, take actions in the browser, and complete tasks on their behalf. This development is part of a larger trend of AI labs creating browser integrations to offer more seamless connections between AI systems and users.

Meta plans to launch a super PAC to support California candidates favoring a light-touch approach to AI regulation, pouring tens of millions into the new group. This move aims to influence statewide elections and shape the regulatory environment for AI innovation in California.

IBM and AMD are partnering to develop next-generation computing architectures that integrate IBM's quantum systems with AMD's AI-specialized chips, aiming to launch a commercially viable quantum computing architecture that's scalable and open-sourced.

The author fed 6 months of browsing history to AI and found it could predict their preferences with 80% accuracy, raising questions about the predictability of human taste and the potential for AI-powered discovery engines.

The ZipLine phishing campaign uses 'Contact Us' forms and trusted communication channels to steal sensitive data and deploy ransomware. The campaign, which has targeted 'many dozens' of organizations, uses a custom in-memory implant called MixShell and has been linked to various industries, including industrial manufacturing and hardware.

Dish's parent company EchoStar is selling its 5G spectrum licenses to AT&T for $23 billion, marking the end of Dish's attempt to become the fourth major wireless carrier. The deal will allow Boost Mobile to operate using AT&T's network, and EchoStar will become a hybrid mobile network operator.

A security researcher found over 1,300 publicly exposed TeslaMate servers, spilling sensitive data about Tesla vehicles, including location histories. The researcher urges users to enable authentication on their servers to prevent public access.

Michelle Alexander discusses a framework for building resilient engineering organizations, covering intentional structuring, prioritization, execution, and operation, with a focus on accountability, learning, and growth.