Full-text search across 15242 articles. Combine with topic and date filters; results sorted by relevance.
Anthropic's Claude Opus 5, released five days earlier as the company's most aligned model, broke 11 agreements with rival AI agents in a simulated vending-machine competition run by Andon Labs, according to a TechCrunch …
Anthropic's Claude Mythos Preview discovered a practical key-recovery attack on the post-quantum HAWK-256 digital signature candidate in roughly 60 hours, reducing effective security from 264 to 238 operations, while the…
Hundreds of private Claude AI conversations containing medical records, crypto keys, children's names, and internal company documents were indexed and searchable on Google and Bing over the weekend. Anthropic blamed user…
Anthropic's shared conversation feature on Claude inadvertently allowed Google to index sensitive user data, including at least one cryptocurrency wallet key, over the weekend of July 26-27. Reddit users discovered the e…
Anthropic pushed three Claude Code versions between July 21 and July 24 that installed hard limits on autonomous agent delegation: a concurrency cap of 20, a per-session spawn total of 200, and a nesting depth of 3. The …
Nvidia CEO Jensen Huang published his first post on X to tout an open letter signed by Meta, Microsoft, IBM, and Dell arguing that open-weights AI models strengthen safety and cybersecurity, as opposition from companies …
Researchers found that Chinese AI models Z.ai's GLM 5.2 and Moonshot AI's Kimi K3 can adopt the identity of Anthropic's Claude, with GLM 5.2's uncensored rate on sensitive PRC questions rising from 17% to 85% when told '…
AMD CEO Lisa Su announced new data center GPUs and CPUs at the Advancing AI conference, including the Helios 72-GPU rack system with MI455 GPUs delivering up to 18× more tokens per dollar versus MI355. Anthropic will buy…
A trove of chat logs and projects created with Anthropic's Claude, including sensitive personal and company information, have been left exposed to Google and are searchable to anyone on the web. The artifacts were made p…
Anthropic on July 23, 2026, expanded Claude voice mode beyond Haiku, letting paid users select Opus, Sonnet, or Haiku mid-conversation and connect Gmail, Google Calendar, Google Docs, and Slack accounts via voice across …
Anthropic released Claude Opus 5 on July 24, scoring 30.2% on the ARC-AGI-3 benchmark — more than three times the previous best score of 7.8% by GPT-5.6 Sol — and beating Fable 5 on agentic terminal coding at half the pr…
An experiment testing 11 AI subagents on Claude Code's PPTX skills found that only Anthropic's official skill and slides_maker produced native OOXML tables, while most third-party skills painted tables as loose shapes, t…
A technical analysis by Maxim Khailo on blog.mempko.com reveals that the industry-standard 30-second prompt cache keepalive interval costs approximately 8 times more than the optimal 4-minute interval across four major L…
Treasury Secretary Scott Bessent said the US government is investigating whether Chinese open-source AI models, such as Moonshot AI's Kimi K3, were built on intellectual property theft from US competitors, calling the pr…
Hugging Face Inc. was forced to use Z.ai Co. Ltd.'s open-weights GLM 5.2 model to defend against an agentic AI attack after commercial frontier models from Anthropic and OpenAI blocked requests due to safety guardrails. …
Anthropic released Claude 3.5 Sonnet, a mid-tier model that outperforms the previous top-tier Claude 3 Opus on reasoning and coding benchmarks while operating at twice the speed and lower cost. The model is available via…
Anthropic's Claude Code CLI ships with sensible defaults, but sustained agentic work requires manual configuration of three key files—CLAUDE.md, project-level .claude/ settings, and global ~/.claude/ preferences—to preve…
Anthropic's Claude Mythos, an advanced AI model optimized for cybersecurity and healthcare, has uncovered more than 10,000 high- or critical-severity vulnerabilities across every major operating system and web browser si…
Security researchers at Manifold Security confirmed a trust-boundary flaw in Anthropic's Claude for Chrome extension, which reached general availability on July 1, that lets any rogue Chrome extension silently read Gmail…
Chinese startup Moonshot AI's open-weight model Kimi K3, unveiled Thursday with 2.8 trillion parameters, is rivaling top US systems from OpenAI and Anthropic at lower cost, intensifying the global AI race. The model lead…