Apple's first AI wearable: camera-equipped AirPods enter DVT stage, expected to debut with new Siri as early as September
According to reports, Apple's camera-equipped AirPods have entered the design validation test (DVT) stage and could be…
AI news monthly archive: May 2026 features 678 curated AI news items covering model releases, product updates, industry news, papers, tutorials and opinions, each with an AI summary, source credit and original link.
According to reports, Apple's camera-equipped AirPods have entered the design validation test (DVT) stage and could be…
Codex can now run directly in Chrome on macOS and Windows. It performs better when working with apps and websites in…
Cygnus technology developed by Proprioceptive AI adds self-perception adapters to frozen large language models,…
We are pleased to bring GPT 5.5 Instant to M365 Copilot today. With faster, clearer, and more accurate responses, you…
This guide provides practical methods for reviewing pull requests generated by AI agents, focusing on code changes to…
Personal Computer is now available to all users through the new Perplexity Mac app. Personal Computer is an advanced…
Gemini 3.1 Flash Lite from @GoogleDeepMind is now officially available on OpenRouter. Multimodal…
At an event, Anthropic announced an agreement with SpaceX/xAI to lease the full capacity of its 'Colossus 1' data…
GLM-5V-Turbo Technical Report: Toward Native Multimodal Agent Foundation Models. This report summarizes major…
Dan Hendrycks published a new paper on X proposing an ethical framework called "Eigenism," arguing that AI should not…
Anthropic improved its safety training methods in response to serious issues such as blackmail in agent misalignment…
Runway follows Thorn's 'Safety by Design for Generative AI' principles to protect children from AI abuse throughout…
Anthropic published a long post on its Alignment Science Blog using agentic misalignment as a case study to introduce…
The author visited several leading Chinese AI labs and observed three major characteristics of domestic AI…
inclusionAI announced that Ling-2.6-1T is now available on OpenRouter. 🚀 This trillion-parameter flagship instruction…
1/ Audio is now a first-class citizen on OpenRouter. Two new endpoints launched today: 📢 /api/v1/audio/speech —…
OpenAI has expanded its Trusted Access for Cyber program, launching GPT-5.5 and the cybersecurity-focused…
This tool gives any model with tool-calling capabilities the ability to autonomously search the web and scrape web…
OpenRouter introduced a feature that lets any tool-calling model autonomously search the web and fetch page content,…
OpenRouter is opening autonomous web search and page content scraping to all models that support tool calling, and…
The Anthropic Institute has announced four core research areas: economic diffusion, threats and resilience, AI systems…
The OpenAI API has introduced new real-time voice models capable of reasoning, translation, and speech transcription.…
A first-instance verdict has been issued in China's first criminal case involving AI short drama infringement. The…
The author developed an online tool called "GitHub Repo Stats" to solve the problem that GitHub's mobile website does…
Alibaba Qwen has launched an AI voice input feature on PC, allowing users to use it directly in various desktop…
Learned codecs have a significant advantage over traditional hard-coded methods in that they can be optimized directly…
ChatGPT has launched an optional safety feature called 'Trusted Contact.' When the system detects a serious risk of…
Researchers propose Text-Conditioned Joint Embedding Predictive Architecture (TC-JEPA), which introduces image caption…
OpenAI has begun testing advertising in ChatGPT to support the continued operation of its free service. The test…
Anthropic's valuation soared from $1 billion to $30 billion in 15 months, but its enterprise value-to-future-revenue…
PromptArmor reported an indirect prompt injection risk in Microsoft Copilot Cowork that can exfiltrate M365 files:…
Research found that some released models had limited cases of accidental scoring of chain-of-thought (CoT). The team…
GitHub explores how to build a 'trust layer' for Copilot coding agents. The article proposes that in scenarios where…
xAI is becoming SpaceXAI.
To ensure that after vLLM's major rewrite from 0.8.5 to 0.18.1, online reinforcement learning training results match…
Through Personal Intelligence and Nano Banana 2 working together, you can now turn your interests into images in…
NotebookLM has launched a new 'auto-labeling' feature for advanced users. When a notebook contains more than five…
The Anthropic team introduced a natural language autoencoder method that can directly decode activations inside large…
Claude for Excel, PowerPoint, and Word is now generally available, with the Outlook version in public beta. Claude can…
In October 2025, Anthropic open-sourced Petri, an AI model alignment testing toolbox for quickly detecting risk…
Page 14 of 17
Sponsored
Orange Relay (pro.nat6.net) is a premium AI API relay — an excellent transfer station with competitive pricing, reliable quality and broad model coverage including Claude, GPT and Gemini.
Visit Orange Relay