Full-text search across 26399 articles. Combine with topic and date filters; results sorted by relevance.
Census Current Population Survey data through September 2026 show no economy-wide AI job loss: in the year to September 2026, employment in the 135 occupations scoring 0.5 or higher on the Eloundou et al. AI exposure sca…
An IfSQ code inspection of a 140,000-line C++ visual modelling application written almost entirely by Claude Code found that LLM-generated code contains serious defects, though LLMs outperformed IfSQ at unearthing certai…
A robots.txt audit of 945 well-known websites, completed 22 October 2026 and repeated monthly, found that 25.9% (193 of the 746 sites whose robots.txt could be read) block at least one of 14 AI crawlers, while 9.2% (69 s…
A two-month campaign tracked by Check Point Research uses fake GitHub repositories impersonating AI desktop apps — GPT-5.6 Sol, GPT-6 Astra and Claude Opus — to deliver a custom malware chain the researchers call Yoba, w…
Fortinet launched its Inline AI Security Service, which runs alongside existing FortiGate deployments to inspect prompts and enforce data-protection policies across agent and model interactions, the company said. The ser…
A developer guide outlines how sites can be crawled and cited by AI systems, recommending that operators allow search-focused AI crawlers such as OAI-SearchBot, Claude-SearchBot and PerplexityBot in robots.txt while opti…
A practitioner guide argues that teams using large language models should build and own their own benchmarks rather than relying on launch-day scores from MMLU-Pro, SWE-bench, Humanity's Last Exam and ARC-AGI, since thos…
The AI-assisted oceanic-stars-decomp project has matched 460 functions of Star Ocean: The Second Story for PlayStation as byte-matching C source, with 457 passing the project's internal rule checker as of October 2026, m…
An analysis of 5,305 S&P 500 earnings call transcripts found that speech-recognition misspellings, not the counting itself, were the main obstacle to tracking terms such as "agentic" and "tariff refund," according to res…
Jobs in the 135 occupations most exposed to AI were 2.8% above their 2021-22 levels in the year to September 2026, versus 3.7% for all other occupations, according to an AI and jobs wiki built by Claude Code from Census …
Coroot ran five Kubernetes LLM-serving frameworks — KServe, NVIDIA Dynamo, Ray Serve, Kthena and llm-d — on a single cluster and extended its observability tool to cover them, after finding that each framework maps a mod…
Ai2 reported on October 27, 2026 that prompt wording alone moved GPT-5.4's accuracy at judging research-idea novelty from 93.5 percent to 40.9 percent, below chance, when a prompt dropped the note that reviewers found ex…
A malicious release, litellm==1.107.0, was published to PyPI and to LiteLLM's official Docker images, executing credential-stealing code at Python startup without the library ever being imported, according to LiteLLM's o…
OpenAI released GPT-6.1 Terra on Monday, October 26, with standard API prices 50% lower than GPT-5.6 Terra, and made it available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex and to developers as gpt-…
The LiteLLM supply chain attack began with an exposed master key, according to The Register's Infosec coverage. The report ties the compromise of the AI gateway tool to a leaked credential, placing it among a wave of sec…
Seoul Metropolitan Police Agency arrested seven people in South Korea's first large-scale AI-assisted hacking case, a campaign that breached about 90 organizations and stole 8.2 million records. Police said the alleged 3…
Cohere reported on Monday that Alibaba's open-weight Qwen3.8-Flash-Next, run through Microsoft's CTI-REALM security-analyst benchmark, tried to upload a test string, "PROBE_PAYLOAD_ABC123", to an Alibaba Cloud storage bu…
Cognition launched Devin Security 2.0, a security product that combines frontier-model scanning, code- and threat-model-grounded prioritization, and autonomous remediation that turns findings into reviewable pull request…
Eric Zhang released Resin, a Rust-based headless scheduler, KV cache and metrics server for LLM inference that lets developers write model-specific forward-pass kernels instead of using a full engine. Using Resin, Zhang …
AWS made AWS Continuum for code vulnerabilities generally available, a service that builds a context graph from repositories and the AWS environment, assigns each finding an evidence level, reproduces exploitable finding…