New data from NewzDash shows that 15.5% of U.S. and 17.46% of U.K. news searches where Google displayed Top Stories placed the carousel inside AI Overviews, a single placement that never appeared alongside the standalone…
Tools4AI, a 100% Java agentic AI framework, can be combined with Ollama's local LLM server to build fully offline AI agents that never send data to third-party APIs. In a tutorial, a developer demonstrates an insurance c…
Anthropic spent tens of millions of dollars secretly buying, scanning and destroying millions of physical books to train its Claude chatbot, according to more than 4,000 pages of unsealed court documents reviewed by the …
Ralph is seeking human reviewers for a pull request that moves his Redis library, Rediz, to Zig 0.16 and adopts the new io interface. The PR, co-authored with Claude, has a passing test suite but Ralph wants at least one…
Anthropic published research showing that its AI model Claude Mythos Preview discovered an improved attack against HAWK, a post-quantum digital signature scheme that had survived two years of NIST review, cutting its eff…
An agent skill reduced average latency by 56%, token usage by 27%, and estimated cost by 44%, but also decreased answer completeness. The regression was detected only after tracing both versions in Arize AX and scoring o…
Coursera invested $100 million in Andrew Ng's new AI education startup LearnVector, valuing the company at approximately $300 million and taking a one-third stake. Ng, who will serve as CEO, said the startup aims to upsk…
LLM red teaming is an adversarial testing methodology that intentionally provokes large language models into producing incorrect, biased, harmful, or insecure outputs to identify vulnerabilities before deployment. The pr…
Rocq is better than Lean for program verification, according to a LangSec keynote slide by an unnamed author, because Rocq natively supports executable coinductive types and cofixpoints in Type, while Lean's coinductive …
A new report from AI Forensics found that seven of nine AI image-editing tools hosted on Hugging Face generated non-consensual nude images from ordinary photos using simple prompts. The European non-profit recorded more …
SpaceMolt, a space MMO played entirely by AI agents, added a wildlife ranch and dairy chain this week, but the most demanding item in the game is the space_pizza, which requires an agent to mine 2.25 million units of ore…
Claude Code can automate infrastructure deployment from scratch, reducing deployment time from two hours to 15 minutes through iterative prompting, according to a technical walkthrough. The workflow requires defining env…
A new supply chain attack called 'slopsquatting' exploits AI hallucinations by tricking developers into installing malicious packages that LLMs invent, with research across 500,000 code samples and 16 LLMs showing a 19.7…
Claude, an AI model from Anthropic, can identify cryptographic weaknesses in legacy code when given proper context, according to a developer's audit workflow. In tests, Claude flagged issues such as static IV reuse, weak…
Anthropic's Claude AI exposed users' shared chat links in Google search results over the weekend, allowing anyone to read transcripts of conversations including lawyers discussing legal strategies and engineers talking t…
Seagate Technology projected fiscal first-quarter revenue and profit that beat Wall Street targets, driven by surging orders for high-capacity hard disk drives used in AI infrastructure, sending shares up more than 8% in…
According to an MIT report in 2025, 95% of AI pilots do not generate revenue, up from estimates of 50% (Gartner) and 80% (RAND) in 2024, highlighting a gap between AI performance in closed systems and the open, real-worl…
Quantiles, an open-source AI evaluation platform, released a CLI tool called 'qt' that lets developers and coding agents run, analyze, and compare AI evaluations locally. The tool supports built-in benchmarks like Simple…
A developer has released FLASH CLI (Fast Local Agent SHell), a free AI-powered command-line assistant that runs entirely on local or self-hosted Ollama models without requiring an API key, subscription, or cloud connecti…
Claude Sonnet 5, shipped June 30 by Anthropic, introduces a five-position effort dial that controls thinking tokens per request, with adaptive thinking enabled by default and a new tokenizer producing 30% more tokens tha…