All AI News

Universities including Rikkyo University are warning users not to input personal information into Google’s AI Mode search, as data may be stored and used for AI training. Experts caution this constitutes a silent data leak with long-term privacy implications.

Claude Opus 4.7's updated tokenizer increases token counts by up to 1.46x for text and 3.01x for images, effectively raising pricing costs by 40% despite unchanged rates. This shift impacts enterprise AI budgets and highlights growing tokenization complexity in generative AI.

PrfaaS, a groundbreaking cross-datacenter KVCache architecture, redefines how large language models are served at scale by decoupling prefill and decode phases across geographically distributed infrastructure. Developed by Moonshot AI and Tsinghua researchers, the innovation overcomes traditional RDMA bottlenecks.

AI accountability practices are being actively implemented by federal engineers to ensure transparent, ethical deployment of artificial intelligence systems. Two case studies from the AI World Government event reveal how agencies are embedding responsibility into AI development lifecycles.

Headless AI services are revolutionizing enterprise software by eliminating graphical interfaces in favor of direct API access. Salesforce’s new Headless 360 initiative exemplifies this shift, enabling AI agents to interact seamlessly with core business systems.

OpenMythos is an open-source PyTorch reconstruction of Anthropic's unpublicized Claude Mythos model, achieving performance parity with a 1.3B transformer using just 770M parameters. The project, led by researcher Kye Gomez, offers the first principled architectural blueprint of the elusive model.

TabPFN leverages in-context learning to achieve superior accuracy on tabular datasets, outperforming traditional models like Random Forest and CatBoost. This breakthrough challenges long-standing norms in machine learning for structured data.

A new AI-powered file type detection and security analysis pipeline combines Magika’s byte-level classification with OpenAI’s language intelligence to enhance cybersecurity posture. This innovation reduces reliance on filenames and enables deeper threat identification.

Uber is entering its assetmaxxing era, leveraging AI to maximize efficiency across its transportation network. This strategic shift redefines urban mobility by reducing reliance on physical assets and enhancing algorithmic control.

A new open-source framework called Proxy-Pointer RAG claims to achieve 100% accuracy in retrieval-augmented generation by merging structured and vector search. The system promises a five-minute setup and represents a significant leap in making RAG systems more reliable. This development could address critical hallucination issues plaguing current AI implementations.

Anthropic's new Claude Design tool has sent Figma's stock tumbling, but the launch is overshadowed by widespread user complaints over billing errors, login failures, and hidden usage drains. Claude Design is here — and the fallout is just beginning.

A new AI coding contest pits Claude against Gemini in solving a complex, weighted variant of the classic knight's tour problem. The 'laden knight's tour' challenge tests the models' algorithmic reasoning and optimization capabilities. This head-to-head competition highlights the evolving role of AI in advanced problem-solving.

A user investigation reveals that ChatGPT's notorious slowdown in long conversations is not a server-side issue but a client-side browser memory problem. The web interface renders every past message simultaneously, causing performance to degrade as chat history grows. This discovery has sparked discussion among developers seeking solutions.

AI vendors are increasingly dismissing critical security vulnerabilities as 'working as intended,' shifting blame to users rather than addressing systemic flaws. This pattern reveals a troubling lack of accountability in the AI industry.

Google’s new TurboQuant framework revolutionizes KV cache management by compressing attention memory with near-lossless precision, enabling massive context windows without prohibitive VRAM costs.

AI Books Feature transforms reading into immersive roleplaying experiences, letting users interact with characters from classic literature. The innovation by Character.AI comes amid growing scrutiny over its past association with teen mental health issues.

Opus 4.7's new tokenizer increases token usage by up to 47% compared to 4.6, driving up costs for users despite Anthropic's unchanged per-token pricing. This shift impacts Claude Code and enterprise users relying on predictable AI spending.

Despite identical per-token pricing on paper, Anthropic’s Opus 4.7 is driving up costs for users due to a new tokenizer that inflates token counts by up to 47%. This hidden cost surge is reshaping how developers budget for AI inference.

A humanoid robot has outpaced human runners in a Beijing half-marathon, setting a new benchmark in robotics. The achievement underscores rapid advances in AI-driven mobility and China’s growing leadership in humanoid technology.

AI-generated influencers are flooding TikTok, Instagram, and YouTube with pro-Trump content ahead of the midterms, amassing millions of views and even catching the attention of Donald Trump himself. Experts warn of growing risks to democratic discourse.

NVIDIA has launched Ising, the first open quantum AI model family designed for hybrid quantum-classical systems, bridging the gap between lab-scale quantum hardware and real-world applications. The move comes as quantum computing gains momentum, with indirect industry impacts including billionaire wealth creation at rival firms.

To maximize Claude Code's potential, developers must structure their projects with engineering discipline. Without clear organization, the tool's inconsistent outputs and context loss undermine productivity.

A new study reveals that always-on Ray-Ban Meta glasses integrated with OpenClaw AI agents significantly accelerate routine tasks by continuously interpreting environmental and digital inputs. The research highlights transformative gains in productivity and contextual awareness.

Agora Labs has launched ABot, the world’s first full-stack embodied AI system designed for AGI, achieving 15 state-of-the-art benchmarks. This breakthrough merges advanced reasoning models with self-evolving memory architecture to create a closed-loop learning ecosystem.