A research implementation called PBTune applies Population-Based Training (PBT), an evolutionary algorithm from DeepMind, to automatically tune PostgreSQL database configurations without machine learn…
Microsoft launched MAI-Cyber-1-Flash, its first cybersecurity-specialized AI model, and a new agentic cybersecurity platform called Perception at an event in San Francisco. Mustafa Suleyman, CEO of Mi…
Wiz Research has built Atlas, an autonomous AI system for vulnerability research that ranks #1 on the CyberGym benchmark with a 90.9% success rate and has uncovered more than 200 previously unknown vu…
Enigma, a physical AI startup founded by Unit 8200 veterans Jonathan Jacobi and Gal Niv, emerged from stealth with a $71 million seed round led by Index Ventures and Ribbit Capital to build an intelli…
Building artificial general intelligence (AGI) via reinforcement learning (RL) and model-based search is terrifying because such algorithms ruthlessly maximize a reward function written in Python, whi…
A new book by Cambridge researcher Eleanor Drage, 'What If We Got AI Right?', argues that the public must look past Silicon Valley's narratives to see the human labor and environmental costs of AI, bu…
OpenAI CEO Sam Altman declared on the 'Relentless' podcast that humanity has entered the singularity, the point at which artificial intelligence surpasses human intelligence and advances unpredictably…
Australia should not try to win the AI arms race but instead leverage its critical minerals and renewable energy while ensuring global tech giants pay fair tax, argues AirTree co-founder Daniel Petre.…
Open weights AI models, predominantly from China, threaten the dominance of frontier labs like OpenAI and Anthropic, which have grown revenues rapidly but may be disrupted before capitalizing on their…
OpenAI CEO Sam Altman said on the 'Relentless' podcast that the world has reached the technological singularity, the point where artificial intelligence surpasses human intelligence. 'We are now, like…
A new personal preparedness guide argues that AI could dramatically upend the world within 2-4 years, citing forecasts from those closest to the technology. The author recommends actions such as helpi…
Speculative decoding can speed up local large language model inference by 1.5 to 2.5 times without changing output quality, according to research from Google and DeepMind. The technique uses a small d…
Anthropic released Claude Opus 5 on July 24, offering near-Fable-5 performance at the same pricing as Opus 4.8 ($5 per million input tokens, $25 per million output tokens), deliberately cannibalizing …
Microsoft is replacing OpenAI's image-generating models with its own MAI technology in PowerPoint and Bing, citing 85% lower costs, and will phase out SMS-based logins for its Entra ID service by Febr…
A cultural shift toward "vibes-based thinking" is replacing probabilistic and frequentist reasoning in how researchers and experts assess catastrophic risks, according to an analysis of contemporary r…
On July 23rd 2026, Dev (@DevvMandal) announced the release of AutoCAD-Bench, a benchmark that tests AI models on precise AutoCAD tasks using computer-use alone. GPT 5.6 sol, a proprietary version of O…
Google's Gemini 3.5 Pro model is months behind schedule, and six current and former DeepMind employees told Axios that low morale is slowing releases. The talent exodus includes Gemini co-lead Noam Sh…
Google's data centers are processing seven times more AI tokens than last year, driven by the rise of agentic AI workloads, according to Mark Lohmeyer, vice president and general manager of AI and com…
Alphabet CEO Sundar Pichai said during the Q2 2026 earnings call that Google needs to improve its coding and agentic coding capabilities, and that a larger Gemini 4 base model will be necessary to sta…
Anthropic has donated another $20 million to Public First Action, a political group advocating for AI regulation, bringing its total contributions to $40 million. The company argues that transparency …