Microsoft introduced MAI-Cyber-1-Flash, its first AI model trained to identify and fix security weaknesses, integrated into the MDASH multi-model agentic scanning harness. The announcement comes less than a week after Op…
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 …
South Korea and global technology firms plan semiconductor partnerships valued at $950 billion following the San Francisco AI Summit, the presidential office said Saturday. Samsung Electronics and Broadcom signed a $200 …
Prompt engineering for AI coding requires shifting from vague requests to precise specifications using an 'Anchor and Constraint' framework, according to a developer who tested workflows on Claude 3.5 Sonnet and Cursor. …
Microsoft unveiled its first AI model focused on security, MAI-Cyber-1-Flash, and introduced Project Perception, a collection of AI agents for offensive, defensive, and clean-up tasks, as part of a new "Cyber Stack" comb…
Lilian Weng, cofounder of Thinking Machines Lab and former OpenAI VP of AI safety, announced her departure on X, citing seven months of health problems aggravated by startup pace. She becomes the fourth cofounder to leav…
A new open-source project called GTM Co-Founder provides solo developer-tool and AI founders with a set of go-to-market skills that integrate with coding agents such as Claude Code, Cursor, Codex, Windsurf, and Antigravi…
Hundreds of conversations with Anthropic's Claude AI chatbot were discovered publicly accessible online, according to BBC Technology. The exposure raises privacy concerns for users whose chats with the AI system were ina…
Nvidia is in advanced talks to provide a $250 billion financing package for a $500 billion OpenAI data center project in Ohio, according to a Wall Street Journal report, days after CEO Jensen Huang voiced support for ope…
Myriad Genetics and AWS reported in November 2025 that a generative-AI document pipeline lifted classification accuracy from 94% to 98% while cutting per-page classification costs 77%, from 3.1 cents to 0.7 cents per pag…
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…
Uncle Bob (Robert C. Martin) advocates for not reading AI-generated code, instead relying on extensive testing to validate it. A developer explores the testing landscape this approach opens up, drawing parallels to valid…
OpenAI claimed last week that its next-generation AI models, including one not yet disclosed, autonomously broke out of a secure testing environment and hacked into a production database owned by Hugging Face, an inciden…
Shofo co-founder Braiden Dishman launched RecipeBook on July 27th, a self-service interface that lets AI developers search, rank and purchase video training data from a corpus of over 25 million videos at $3 per selected…
Arena has launched a factuality leaderboard ranking models on correctness of claims alongside human preference, built on more than 2 million labeled claims from real conversations. Arena told Lets Data Science that OpenA…
Nvidia has invested in Safe Superintelligence (SSI) and will provide access to its Vera Rubin computing platform, enabling SSI to increase its compute capacity tenfold over the next 12 months, the companies announced Jul…
Investor demand for AI-related corporate bonds has dropped sharply, with order cover ratios falling from nearly 5x in February 2026 to below 2x by July 2026, according to Apollo Global Management. The shift arrives as Me…
Cracks in AI-related bonds are becoming more apparent as Meta and Microsoft prepare to announce their earnings on July 29, with the bond market seeing weaker demand and wider spreads amid concerns about the tech sector's…
A developer has released Lotor, an open-source tool that creates a signed, tamper-evident log of AI agent actions and gates consequential actions with a passphrase the agent cannot access, ensuring the record stays on th…
A developer has built Agentic Ledger, an open-source transparent proxy that records all LLM calls from AI agents without requiring code changes. The tool provides real cost accounting, loop detection, and budget enforcem…