Latest AI News and Tool Updates

The New York Times has issued a legal demand to Brave Software, demanding that the browser developer disable features that bypass digital paywalls and terminate arrangements involving the supply of news content to artificial intelligence companies. The publisher contends that Brave unauthorizedly scrapes copyrighted articles to power AI search summarization and provide data to third-party AI developers without licensing agreements. The dispute underscores expanding legal friction between digital publishers and tech platforms over content scraping and AI model training. Brave maintains that its privacy features and content-blocking tools operate locally on user devices within established legal boundaries.

A GitHub issue report highlights user concerns over Postman automatically installing AI companion features and enabling default settings that transmit API data to artificial intelligence models. Community members report that these automatic integrations risk exposing sensitive API payloads, authorization headers, and confidential tokens to external LLMs without explicit user consent. Developers and enterprise administrators argue that opt-out AI deployments violate corporate security compliance and demand strict opt-in controls, transparent data governance policies, and simple administrative toggles to completely disable AI integrations across teams.

Axel Scheffler, illustrator of The Gruffalo, is considering legal action against AI companies over the unauthorized use of his artwork in model training. Testing of popular AI tools revealed that ChatGPT readily generated near-exact replicas of the Gruffalo monster when prompted directly, indicating its image generator was likely trained on copyrighted images. While Google Gemini refused direct character requests due to copyright guardrails, it successfully emulated Scheffler's signature watercolor illustration style when prompted. These findings highlight ongoing copyright and intellectual property concerns as artists push for legal protections against style replication and unauthorized data scraping.

A growing majority of recruiters believe job applicants no longer write their own CVs, relying instead on generative AI to draft and tailor resumes to fit job descriptions. According to recent research, candidates increasingly use AI tools to match recruiter expectations and bypass automated applicant tracking systems. While AI assists job seekers in structuring qualifications quickly, hiring managers report encountering generic, inflated, or homogenized profiles. Consequently, recruiters are placing greater emphasis on live interviews, practical assessments, and verification steps to evaluate a candidate's true skills and authentic experience beyond AI-crafted applications.

Proton CEO Andy Yen discusses the privacy risks and centralization challenges posed by rapid artificial intelligence adoption, warning against Big Tech's dominance over personal data and AI infrastructure. Highlighting Proton's privacy-first strategy, Yen outlines the deployment of privacy-preserving tools like local AI writing assistant Proton Scribe, which processes user requests on-device without data logging. He further explains Proton's structural transition to a non-profit foundation designed to protect its encrypted ecosystem and ensure long-term independence from commercial pressures.

Best Buy has discounted the Minisforum AI X1 Pro-470 mini PC, undercutting rival retailers like Amazon. Designed as a compact Copilot-ready workstation, the mini PC integrates a dedicated Neural Processing Unit (NPU) for handling on-device AI tasks alongside standard desktop workflows. The system features strong expandability options, including user-upgradable RAM, dual storage slots, and an OCuLink port for attaching external GPUs. For users seeking an accessible compact desktop for local machine learning experiments or intensive daily workloads, the price cut offers a versatile workstation setup at a lower cost.

Netflix has introduced an AI-powered Discover tab that recommends movies and TV shows based on a user's current mood and natural language prompts. Rather than relying solely on traditional genre rows, subscribers can enter custom descriptions or select emotional themes to receive tailored recommendations from the platform's catalog. Powered by generative AI models that process content metadata and viewing history, the feature aims to streamline content selection and reduce decision fatigue for subscribers searching for specific viewing experiences.

Meta is ending new software feature updates for the Quest 2 and Quest Pro VR headsets, transitioning the older hardware into maintenance mode. While critical security patches will continue for a limited timeframe, new OS enhancements and developer APIs will no longer be delivered to these devices. This shift allows Meta to focus software development on its newer Quest 3 series and its upcoming AR and VR smart glasses, which integrate multimodal Meta AI capabilities and lightweight optics. Existing Quest 2 and Quest Pro owners can continue playing current titles, but will not receive future spatial computing features.

CrowdStrike is acquiring two AI security startups in deals totaling nearly $500 million to address the growing demand for securing autonomous AI agents. As businesses increasingly deploy AI agents to execute automated tasks, organizations face novel security risks such as prompt injection, data leakage, and unauthorized system actions. The planned acquisitions will integrate specialized threat detection and governance tools into CrowdStrike's security platform, allowing enterprises to monitor agent behavior, enforce access controls, and protect agentic workflows against emerging cybersecurity vulnerabilities.

To determine whether software improvements alone could surpass 1990s supercomputing, Nicholas Carlini evaluated modern chess engines constrained to 1996 Pentium hardware limits. IBM's Deep Blue defeated world champion Garry Kasparov in 1997 using custom hardware capable of evaluating 200 million positions per second. However, three decades of algorithmic advances—including refined search heuristics and neural network evaluation functions—have gained over 1,000 Elo points independent of raw processing power. Running on a single 1996 Pentium CPU evaluating only thousands of positions per second, modern Stockfish comfortably exceeds Deep Blue’s playing strength and would have defeated Kasparov in 1996.

Anthropic is spending more than $100,000 per employee per month on compensation as competition for top artificial intelligence talent escalates across Silicon Valley. This figure, which encompasses base salaries and substantial equity packages, highlights the immense financial pressure on AI frontier labs competing directly against deep-pocketed tech giants like Google and Meta. As companies race to build increasingly sophisticated foundation models, the limited pool of specialized AI researchers and engineers has driven compensation packages to historic highs. Alongside soaring infrastructure and compute costs, these talent expenses contribute significantly to Anthropic's rising burn rate as it seeks to maintain its competitive edge against rivals like OpenAI.

TechRadar's Black Friday guide highlights top deals on Bambu Lab 3D printers, featuring recommendations based on expert hands-on testing. The entry-level Bambu Lab A1 Mini is highlighted as a compact, budget-friendly option for beginners, while the standard A1 offers a larger build volume and multi-color printing capabilities through the AMS Lite system. For enthusiasts and professionals, the flagship X1-Carbon features an enclosed design for technical filaments, fast print speeds, and AI-driven camera monitoring for automatic failure detection.

Tesla has updated the name of its driver-assistance software in the UK and Europe to "Full Self-Driving (Supervised)" following regulatory scrutiny over potentially misleading claims. Despite the moniker, the feature remains a Level 2 driver-assist system, requiring human drivers to stay alert and ready to take control at all times. The rebranding aligns with stricter European advertising standards aimed at preventing confusion regarding vehicle capabilities. Meanwhile, rival automakers including BMW are continuing development on advanced automated driving systems, with plans to introduce higher-level autonomous features in production vehicles by 2027.

Google is providing K-12 school administrators with administrative controls to turn off standalone AI services like Gemini for students using Google Workspace for Education accounts. However, because AI Overviews are directly embedded into Google Search, schools cannot turn off these AI-generated search responses without disabling or restricting access to Search altogether. This integration forces educational institutions into an all-or-nothing choice between exposing students to AI features or cutting off standard search access.

NVIDIA has open-sourced Auto-Ontology, a framework designed to automate the creation, extension, and refinement of domain-specific ontologies and knowledge graphs from unstructured text using large language models. The tool streamlines structured knowledge extraction by automatically identifying entities, relationships, and taxonomies, reducing the manual labor traditionally required in ontology engineering. By converting unstructured enterprise data into grounded, machine-readable knowledge structures, Auto-Ontology helps developers enhance retrieval-augmented generation pipelines, reasoning systems, and graph-based AI applications.