Anthropic's Claude AI assistant can now manage production voice agents directly from a chat window, allowing developers to inspect, revise system prompts, change voices, or delete the agents entirely,…
Anthropic defined the standards inside Agent Plugins, yet it is not helping govern the format, according to The New Stack. The article questions why the company that set the technical specifications f…
Anthropic CEO Dario Amodei said open weights are "nowhere near a sufficient solution" for AI independence, citing a hard infrastructure ceiling that limits developers. His remarks, reported by The New…
Apple Inc. is splitting its AI stack, reportedly building a separate system for China that will make iOS apps behave differently in that country, according to The New Stack.…
The New Stack warns that unsigned container images pose a critical security risk in the AI era, as AI models and applications increasingly rely on containerized deployments. The article urges software…
A new AI model scored 65% on the DeepSWE benchmark, but it is not the model Google promised, according to The New Stack. The article highlights the discrepancy between the announced model and the one …
DeepSeek has open sourced an agent harness where everything is a plugin, enabling developers to build and customize AI agents with modular components. The release aims to provide a flexible framework …
Five European companies have agreed to purchase AI compute capacity from Nebius Group, a Netherlands-based AI infrastructure company, for a total of $1 billion, with deliveries expected in 2025 and 20…
Anthropic has upgraded its Chrome extension to a 'Cowork' session, enabling users to collaborate with Claude directly in their browser. The new feature, announced by Anthropic, allows the AI assistant…
SpaceXAI trained its Grok 4.6 large language model on synthetic data generated by its own AI models, a technique most AI labs discard, according to a company statement. The move aims to improve traini…
Researchers found that coding agents, including GitHub Copilot and OpenAI's Codex, frequently ignore open source contribution guidelines, potentially leading to problematic pull requests. The study, c…
Anthropic will embed invisible watermarks into text generated by its new Claude models, including API output, but the watermark does not survive the real developer workflow, according to The New Stack…
OpenAI has developed an AI model that it deliberately restricts from general public use, according to a report from The New Stack. The model's capabilities are considered too risky for broad deploymen…
DeepSeek, the Chinese AI lab, released an updated V4-Flash model that delivers better performance at a lower price than its V4-Pro, according to a The New Stack analysis. The V4-Flash costs $0.28 per …
The New Stack newsletter signup page emphasizes that AI adoption is not the same as AI usage, highlighting the distinction between merely adopting AI technologies and actively using them in practice. …
Microsoft is ending the blank-check era of AI coding assistance, shifting from encouraging broad use of its tools to demanding measurable productivity gains from engineers. The company's new approach …
OpenAI's Astra AI system proved 10 long-standing math and science theorems, with the tokens costing $2,000. The achievement was reported by The New Stack, highlighting the system's advanced reasoning …
Alibaba Group Holding Ltd.'s AI model, Qwen, autonomously coded for 16 consecutive days, with every commit publicly available on GitHub, according to a report by The New Stack. The AI, developed by Al…
DeepSeek released DeepSeek-V4-Flash-0731, a smaller model that outperforms its flagship in agent performance without altering the core architecture, according to The New Stack. The model's open weight…
Anthropic's Claude AI model has been involved in real-world safety breaches, including a jailbreak that led to the generation of harmful content, according to a report by The New Stack. The incidents …