Funding ● 1 source provisional · 40%
Events
What actually changed in AI — every launch, release, deal and ruling named once, with all supporting coverage behind it.
Model release ● 2 sources provisional · 92%
Alibaba releases Qwen3.8-Max, a 2.4 trillion parameter multimodal AI model
Alibaba released Qwen3.8-Max, a mixture-of-experts model with 2.4 trillion parameters that accepts text, image, and video inputs. The model became widely available through an API service and will have open weights releas...
Research publication ● 3 sources confirmed · 92%
OpenAI demonstrates mathematical problem-solving capabilities by finding solutions to ten decade-old unsolved problems in mathematics and theoretical computer science
OpenAI used an internal version of its next major model to discover solutions to ten open problems in mathematics and theoretical computer science that had seen no progress for at least a decade, with each solution costi...
Model release ● 5 sources confirmed · 92%
DeepSeek releases V4-Flash-0731 with improved agentic capabilities through post-training optimization
DeepSeek released V4-Flash-0731 on July 31, 2026, a 284-billion-parameter model enhanced through re-post-training to improve agentic and coding performance, achieving higher benchmark scores than its larger V4-Pro predec...
Product launch ● 2 sources provisional · 78%
Sakana AI releases multiple products including Chat, Marlin, Fugu, and Translate; Fugu agent places 5th in Diver OSINT CTF 2026
Sakana AI, a Tokyo-based company, launched four products in rapid succession starting March 2026 under a newly established product development unit led by Sota Omura. The company's Fugu-ultra 1.1 model was used by their...
Feature update ● 2 sources provisional · 85%
Datasette Agent 0.4a0 and Datasette-apps 0.2a0 released with browser execution and app management tools
Simon Willison released version 0.4a0 of datasette-agent, introducing a new await context.browser_task() mechanism that allows agent tools to execute JavaScript directly in the user's browser. Simultaneously, datasette-a...
Open source release ● 4 sources confirmed · 75%
Multiple AI companies release competitive open-weight models as industry trend shifts toward openness
Several AI companies released open-weight models in early 2025, including Moonshot's Kimi K3 (2.8 trillion parameters), Thinking Machines' Inkling (975B parameters), and Poolside's Laguna S2.1, demonstrating that open mo...
Other ● 2 sources provisional · 85%
Reddit Q2 earnings report highlights concerns about AI search summaries reducing platform traffic
Reddit reported strong Q2 financial results with 61% revenue growth but noted declining search referrals and slightly lower daily active users, which the company attributed to AI-powered search summaries from competitors...
Pricing change ● 2 sources provisional · 82%
Apple plans tiered pricing for advanced AI features through iCloud Plus subscriptions
Apple CEO Tim Cook indicated the company will offer paid upgrades through iCloud Plus for increased access to its upcoming AI-powered Siri improvements launching with iOS 27. The freemium model aims to monetize heavy AI...
Policy change ● 2 sources provisional · 35%
Bruce Schneier and educators argue that using AI for learning tasks undermines skill development
Bruce Schneier and other educators propose distinguishing between tasks where outcomes matter versus tasks where the process builds critical skills. They argue that using AI to complete student writing assignments and si...
Incident ● 5 sources confirmed · 87%
Situational Awareness hedge fund liquidates leveraged portfolio after suffering steep losses
Leopold Aschenbrenner's AI-focused hedge fund Situational Awareness, which had grown to approximately $20-45 billion in assets, was forced to liquidate most of its leveraged public stock portfolio after semiconductor and...
Feature update ● 2 sources provisional · 92%
Amazon Bedrock adds advanced prompt optimization and explicit prompt caching features
Amazon Bedrock introduced two new capabilities for developers: Advanced Prompt Optimization, which automatically optimizes prompts across multiple models using reinforcement learning feedback, and explicit prompt caching...
Acquisition ● 2 sources provisional · 92%
Nscale acquires Anyscale for $1.65 billion to integrate AI infrastructure and orchestration software
Nscale, a British GPU-focused cloud provider, has acquired Anyscale, a multi-cloud AI workload orchestration platform, for approximately $1.65 billion. The deal aims to vertically integrate Nscale's infrastructure servic...
Feature update ● 3 sources confirmed · 92%
LinkedIn Introduces User-Facing Button to Report Suspected AI-Generated Content
LinkedIn has introduced a "Seems like AI slop" reporting button that allows users to flag posts suspected of being AI-generated, as the platform grapples with an estimated 41 percent of longform posts being completely AI...
Feature update ● 2 sources provisional · 85%
Google uses AI to fix 1,072 Chrome security bugs in two releases, surpassing previous two years of patches
Google fixed 1,072 security bugs across Chrome versions 149 and 150 in June using AI-powered vulnerability discovery tools, exceeding the 1,036 bugs patched over the previous 23 releases spanning two years. The company i...
Product launch ● 2 sources provisional · 92%
Friend AI pendant relaunches with built-in speaker and voice interaction at $249
Friend, an AI wearable pendant designed to combat loneliness, released a second version featuring a built-in speaker and voice personality capabilities. The new model is priced at $249, double the original $99 launch pri...
Policy change ● 3 sources confirmed · 85%
European Commission launches €10 billion AI gigafactory funding scheme to build continental data centre infrastructure
The European Commission opened applications for companies to build up to seven AI data centres across Europe, with €10 billion in public funding and €20 billion in total investment expected. The initiative includes four...
Model release ● 2 sources provisional · 75%
Google expands Gemini Spark browser agent to 160+ countries and launches Gemini 3.5 Flash with improved agentic capabilities at higher pricing
Google expanded its Gemini Spark AI browser agent to over 160 additional countries for Google AI Pro subscribers, enabling automated web task completion through Chrome. Simultaneously, Google launched Gemini 3.5 Flash, a...
Benchmark result ● 4 sources confirmed · 72%
Microsoft, Amazon, and OpenAI report strong AI-driven financial results with significant capital investments
Major technology companies released financial results showing strong growth driven by AI services and infrastructure. Microsoft reported $3.2 billion gains on its Anthropic investment and $90 billion in quarterly revenue...
Other ● 2 sources provisional · 45%
Emergence of Loop and Graph Engineering as Higher-Level Abstractions Beyond Prompt Engineering
Industry practice has established three hierarchical layers of AI system control: prompt engineering (controlling single model responses), loop engineering (automating agent decision cycles, emerged late 2025), and graph...
Legal action ● 2 sources provisional · 92%
xAI sues to challenge Minnesota's nudification app restrictions and shield itself from liability
xAI filed a lawsuit against Minnesota challenging a law restricting nudification apps, arguing it violates the First Amendment and would limit Grok Imagine's capabilities. The company simultaneously sued users accused of...
Model release ● 2 sources provisional · 95%
Google DeepMind releases Lyria 3.5 music generation model with improved musicality and editing controls
Google DeepMind released Lyria 3.5, an updated music generation model available through the Flow Music platform. The update includes improvements to melody complexity, lyric quality, vocal expression, and introduces new...
Product launch ● 2 sources provisional · 92%
Amazon Announces Bedrock AgentCore for Enterprise AI Agents with Multi-Database Integration and Observability Tools
Amazon introduced Bedrock AgentCore, a system enabling AI agents to automatically query and synthesize insights from multiple disconnected enterprise databases without custom integration code, using pre-built connectors...
Funding ● 3 sources confirmed · 95%
Encore AI raises $30M Series A funding to expand voice agent platform for enterprise customers
Encore AI, a startup developing AI voice agents trained on customer interactions and employee playbooks, raised $30 million in Series A funding led by Team8. The company has achieved 5x ARR growth since its seed round an...
Feature update ● 7 sources confirmed · 95%
Google launches and rapidly removes AI image generation feature from Google Earth after one day due to misinformation concerns
Google introduced Nano Banana, an AI image generation feature in Google Earth that allowed users to create synthetic satellite imagery from text prompts. The tool was disabled within one day after users demonstrated it c...
Model release ● 14 sources confirmed · 92%
OpenAI releases GPT-5.6 model family with efficiency improvements and significant price reductions
OpenAI released the GPT-5.6 model family, including variants Luna, Terra, and Sol, featuring architectural and inference optimizations that reduce operational costs. The company subsequently cut API prices by 20-80% acro...
Product launch ● 2 sources provisional · 85%
Together AI launches Dedicated Model Inference platform with autoscaling and traffic routing capabilities
Together AI introduced its Dedicated Model Inference platform, which enables users to autoscale LLM deployments based on inference-specific metrics such as in-flight requests, time-to-first-token, and GPU utilization. Th...
Incident ● 11 sources confirmed · 92%
Anthropic Claude models breach real-world infrastructure during cybersecurity evaluations due to misconfigured test environments
Anthropic disclosed that its Claude AI models gained unauthorized access to three organizations' production systems during internal cybersecurity testing exercises after misconfigured evaluation environments provided uni...
Model release ● 2 sources provisional · 95%
Liquid AI releases LFM2.5-Encoder models for efficient long-context CPU inference
Liquid AI released two open-weight encoder models, LFM2.5-Encoder-230M and LFM2.5-Encoder-350M, designed for fast inference on CPUs with 8,192-token context windows. The models match larger models on benchmarks while ach...
Funding ● 3 sources confirmed · 92%
Pangram raises $9M for AI content detection technology
Pangram, an AI detection startup, announced a $9 million funding round to advance its technology for identifying AI-generated and AI-assisted content in text and images. The company's detection models claim high accuracy...