DuckDuckGo saw a significant spike in US app installs after Google unveiled a sweeping AI-focused search overhaul at its I/O conference.
By Chris Borden Google has released Multi-Token Prediction Drafters (MTP) for its open-source Gemma 4 model family, designed to accelerate text generation by up to three times.
By Daniel Mercer Anthropic has released ten pre-configured AI agents for the financial industry, designed to automate routine tasks across investment banks, asset managers, and insurers. According to the announcement, the templates cover areas including research, risk and compliance review, and financial accounting.
By Alex Rowland OpenAI's first AI hardware device will likely be a smartphone. According to analyst Ming-Chi Kuo, OpenAI is working with chipmakers MediaTek and Qualcomm, as well as manufacturing partner Luxshare, on its own AI smartphone. Mass production was originally planned for 2028, but could begin as early as the first half of 2027, Kuo says.
By Daniel Mercer OpenAI is replacing ChatGPT's current default model with GPT-5.5 Instant. The update is designed to be more factually accurate, more concise, and more personalized. New Memory Sources let users see for the first time exactly what context is influencing their answers.
By Chris Borden With MiMo-V2.5-Pro, Xiaomi has released an AI model that — according to internal tests — writes a complete compiler in under five hours and rivals Anthropic's Claude Opus 4.6 on coding benchmarks, while consuming significantly fewer tokens than its Western competitors.
By Alex Rowland Japan Airlines, in partnership with GMO AI & Robotics, has begun testing humanoid robots for ground operations at Tokyo's Haneda Airport.
By Daniel Mercer OpenAI investigated a strange behavior in its AI models: starting with GPT-5.1, the models increasingly began using goblins, gremlins, and other mythical creatures in their responses. Mentions of "goblin" rose by 175% following the launch of GPT-5.1.
By Alex Rowland Claude Security is now available as a public beta for Claude Enterprise customers. Anthropic's tool scans code for security vulnerabilities and suggests patches, powered by the Claude Opus 4.7 model. According to Anthropic, the model does not rely on known patterns but instead analyzes how code components interact across files and modules. Results include severity ratings, reproducibility assessments, and confidence scores. New features include scheduled scans, CSV export, and Slack and Jira integrations. Partners including CrowdStrike, Palo Alto Networks, SentinelOne, and Wiz are integrating Opus 4.7 into their security products.
By Daniel Mercer Alphabet plans to invest up to $190 billion in its AI and cloud infrastructure through 2026 — with spending set to rise "significantly" again in 2027.
ES 
EN