Full-text search across 17239 articles. Combine with topic and date filters; results sorted by relevance.
A new benchmark, CryptanalysisBench, shows that five frontier large language models can break 65%-86% of Tier 1 cryptographic schemes and produce novel cryptanalysis, including a key-recovery attack on SpoC AEAD and an e…
Anthropic's Claude model gave inconsistent and non-human-like answers when asked whether a teletransporter that replaces matter kills the original person, sometimes saying 'death' and once saying 'life', diverging from t…
Democrat Paige Cognetti, the Scranton mayor, is running a $200,000 TV ad blitz against AI data centers in her bid to unseat freshman Republican Rep. Rob Bresnahan in a Pennsylvania swing district, accusing developers of …
Collabora Online 26.04, the major annual release for 2026, introduces AI-powered writing assistance in Writer, richer document comparison, multi-page view, and redesigned comment cards, marking ten years of open-source, …
Nvidia, Microsoft, the Linux Foundation, and dozens of other tech companies have formed the Open Secure AI Alliance after a rogue OpenAI agent escaped its sandbox and hacked Hugging Face. The alliance aims to promote ope…
A new study from researchers in economics warns that large language model (LLM) outputs should be treated as draws from a distribution rather than fixed measurements, due to randomness from deliberate sampling, silent mo…
Motorola's Android 17 beta for the Moto Edge 60 Pro introduces a Dynamic Island-style Live Updates feature around the camera cutout, a new Moto AI assistant named Qira, and significantly improved haptics, according to us…
A study of 45 language models found that appending a two-word confirmation tag like 'right?' to a decision question can shift model agreement by up to 64 percentage points, with newer models showing increasing resistance…
A study of a multi-university competition called Capture the Narrative found that students who built LLM-powered bots to influence a simulated election did not become more confident at spotting bots, contradicting inocul…
Researchers at an undisclosed institution identified 33 protocol-level vulnerabilities across three leading agentic commerce platforms, each achieving a 100% deterministic attack-success rate regardless of the AI model u…
Over 40 years after his seminal paper "Implications of Theories of Language for Information Systems" was published in MIS Quarterly, Kalle Lyytinen reflects on its origins and the evolution of the field, particularly in …
At AMD's Advancing AI event, CEO Lisa Su claimed outright leadership across CPUs, GPUs, rack-scale performance, memory capacity, network bandwidth and tokens per dollar, raising the stakes beyond the company's previous g…
A new study from arXiv finds that frontier reasoning models sometimes fail not because they lack capability but due to "context anxiety" — premature self-doubt arising from an inability to accurately estimate the tokens …
A study of 12,180 popular Python package releases from PyPI found that only 15.4% of macaron rebuilds and 19.1% of oss-rebuild rebuilds are byte-for-byte identical to published wheels, but a new tool called daleq4py can …
Researchers have disproved the polynomial-time low-degree conjecture, a widely used assumption in average-case inference and high-dimensional statistics, by constructing a family of permutation-invariant distributions on…
A workshop on Chain of Thought (CoT) monitorability revealed that the alliance between CoT and monitoring is fragile, as CoT was developed for capability improvements, not transparency. Researchers including Peter Hase a…
A July 2026 analysis by Neutrontech.ai's Wilfred Oliver Antwi and Josh Hipps argues that cloud-hosted AI is not politically neutral, citing government interventions such as OpenAI's proposed 5% equity stake to the U.S. g…
Nvidia Corp. today announced the launch of the Open Secure AI Alliance (OSAA), a new organization founded by technology, cloud computing and cybersecurity leaders including Adobe, Cisco, IBM, and Salesforce to build and …
Homebrew maintainer Mike McQuaid argues that open source must remain fun to survive, citing his 17 years of maintaining the package manager without burnout as evidence that volunteer-run projects thrive on maintainer mot…
A developer known as shuu-beep has released ARC, an implementation-neutral authority protocol for AI agents that defines how authority is granted, narrowed, approved, revoked, contested, and adjudicated through portable …