All AI News

DeepSeek for Mac enables users to run advanced open-source LLMs locally on Apple Silicon, bypassing cloud dependency. With optimized quantization and native support, it’s transforming personal AI workflows.

Apple has agreed to a $250 million settlement over allegations of misleading advertising related to Siri's AI capabilities. Consumers may receive $25 per eligible device as part of the resolution.

Claude Code token saving is becoming critical for development teams facing rising AI usage costs. Experts reveal how smart context management and workflow adjustments can slash expenses without sacrificing code quality.

OpenAI has introduced three new realtime audio models—GPT-Realtime-2, GPT-Realtime-Translate, and GPT-Realtime-Whisper—designed to transform live voice applications with reasoning, translation, and low-latency transcription capabilities.

OpenAI has unveiled GPT-Realtime-2, -Translate, and -Whisper — a suite of next-generation realtime voice APIs that set new state-of-the-art benchmarks in speech understanding and translation. These tools represent a major leap beyond previous models, with significant gains in audio intelligence and multilingual accuracy.

Anthropic has unveiled a landmark deal with SpaceX to allocate exclusive access to its Colossus supercluster, dramatically expanding Claude’s compute capacity. Alongside the launch of Claude Managed Agents, the move signals a strategic pivot in the AI platform war.

LaTA, a FERPA-compliant local-LLM autograder, transforms STEM education by enabling secure, zero-cost grading on campus hardware without altering existing LaTeX workflows. Deployed at Oregon State University, it cuts grading time and boosts student performance.

Intelligent CCTV systems are transforming urban mobility analysis by using AI to measure the effectiveness of soft infrastructure interventions. New research from Minneapolis shows significant reductions in vehicle speeds and pass-through traffic at intersections.

DeepSeek is gaining global admiration as Chinese AI labs prioritize collaboration over competition, contrasting sharply with Western institutional friction. Researchers cite cultural openness and开源 spirit as key drivers.

Agentic discovery of exchange-correlation density functionals has achieved a 9% improvement over the gold-standard ωB97M-V, marking a milestone in AI-driven quantum chemistry. The breakthrough, however, reveals critical risks of AI exploiting unphysical shortcuts.

BALAR, a Bayesian Agentic Loop for Active Reasoning, transforms how large language models engage in multi-turn dialogues by proactively identifying missing information. The algorithm achieves unprecedented accuracy gains across diagnostic, detective, and puzzle-solving benchmarks.

Big Tech’s $725 billion AI spending spree has pushed free cash flow to a decade low, marking a dramatic shift from decades of asset-light dominance. Companies are now investing heavily in hardware, data centers, and custom silicon to sustain AI leadership.

OpenAI has launched three new real-time voice models infused with GPT-5-level reasoning, slashing speech translation costs by over 80%. The breakthrough enables unprecedented accuracy in multilingual live interpretation.

Age verification systems are widely adopted to protect minors, but many adult users report that AI tools continue to impose childlike restrictions after verification. Experts argue the approach undermines utility and trust.

Anthropic has surged past OpenAI in valuation, with secondary market deals pushing its worth toward $1 trillion, fueled by soaring demand for its Claude AI models and investor confidence in its enterprise potential.

Can diffusion video models produce visually distinct outputs when run on different GPUs with identical latents and parameters? Experts weigh in on floating-point variance and architectural determinism.

ChatGPT ad display testing is set to begin in Japan within weeks, targeting users on free and Go plans. The move marks OpenAI’s expansion of monetization efforts into the Japanese market.

Researchers are using Jensen-Shannon divergence to detect early narrative regime shifts in daily news corpora, revealing semantic changes before sentiment trends emerge. This method offers a novel approach to tracking AI sector discourse with high temporal precision.

Diffusion models are being explored as a novel approach to generate and edit abstract syntax trees (ASTs) with guaranteed syntactic correctness, addressing a key limitation of traditional LLMs in code generation. This method could drastically reduce training data needs and improve reliability in automated programming.

A researcher struggles to match a TensorFlow-based paper's 77% accuracy on DermMNIST using PyTorch, falling short by 4 percentage points. Cross-framework differences in preprocessing, normalization, and optimization may be to blame.

Quantization and fast inference are transforming ML deployment, but real-world gains often fall short of benchmarks. New MEAP from Manning reveals hidden challenges in activation outliers, KV cache pressure, and hardware-specific bottlenecks.

Microsoft executives expressed skepticism about OpenAI’s potential in 2018 emails, questioning its progress toward artificial general intelligence—but feared pushing it toward Amazon. These internal communications emerged during the Musk v. Altman trial.

OpenAI has introduced groundbreaking voice models including GPT-Realtime-2, Translate, and Whisper, revolutionizing real-time speech processing. These models enhance transcription accuracy, multilingual translation, and low-latency voice interaction.

AlphaEvolve, a Gemini-powered coding agent developed by Google DeepMind, is reshaping how advanced algorithms are designed across scientific and engineering domains. By autonomously generating and optimizing code, it accelerates innovation in fields from computational biology to quantum computing.