Anthropic releases Fable 5.1 and Mythos 5.1, cheaper and less restrictive
Anthropic shipped Fable 5.1, a cheaper and less restrictive version of its top model, alongside Mythos 5.1 for vetted partners.
Read More
Three-Quarters of Americans Oppose Data Centers Near Them, New Poll Finds
A new Heatmap poll shows 75% of Americans don't want a data center built nearby, with opposition growing across every demographic group over the past year.
Read More
Scientists Built a Virtual Town With 10 AI Agents. Five Models, Five Wildly Different Outcomes.
Researchers gave 10 AI agents homes, jobs, and memory in a virtual town. The same rules, five different models, five completely different endings.
Read More
Anthropic Expands Mythos 5 Access to More Defenders, Unveils $35M Open Source Fund
Anthropic broadens defender access to Mythos 5 through partner integrations, an updated Claude Security beta, and a new $35M open source cybersecurity fund.
Read More
Chinese hackers use DeepSeek AI to boost attacks
Taiwanese researcher TeamT5 says Chinese state-linked hackers have more than doubled attack volume after adopting DeepSeek and other open-weight AI models.
Read More
AVO Beats Claude Opus 5 on ARC-AGI-3 by Changing the Harness, Not the Model
A NVIDIA coding agent wrapped around Claude Opus 5 went from 30% to 100% on ARC-AGI-3 by swapping tools, not the underlying model.
Read More
New Mexico Fines Meta $942 Million Over Child Safety Claims
New Mexico has fined Meta $942 million under a public nuisance law, citing child safety failures across its platforms.
Read More
Anthropic’s annual revenue run rate hits $65 billion ahead of fall IPO
Anthropic's annual revenue run rate reached $65 billion by late July, up from about $9 billion at the end of 2025, with a fall IPO planned.
Read More
White House Establishes Framework for Private-Sector Hack-Back Operations
A presidential memorandum directs federal agencies to create a program allowing vetted security companies to conduct approved cyber operations against foreign criminal organizations.
Read More
Grok 4.6 Release Puts xAI Back in the Frontier AI Race
xAI released Grok 4.6 with claims of frontier coding and knowledge work performance, matching GPT-5.6 Sol on key benchmarks while facing reputational headwinds.
Read More
Alibaba ships Qwen3.8-Max at 2.4T parameters, but promised open weights are still missing
Alibaba's 2.4T-parameter Qwen3.8-Max flagship hit general availability on August 3, 2026 with real benchmarks and pricing, but the open-weight release is overdue.
Read More
Mark Zuckerberg’s AI manifesto and the trust problem facing consumer AI
Mark Zuckerberg's 6,500-word essay on personal superintelligence raises questions about trust, tone, and how AI leaders communicate with the public.
Read More
Zhipu AI releases GLM-5.3, claims strongest open-weights coding model
Zhipu AI's GLM-5.3 matches its predecessor's base and gains all come from post-training, with the company calling it the strongest open-weights coding model.
Read More
X open sources its ranking algorithm, letting users see if they have been shadowbanned
X has released the source code for its For You feed ranking system on GitHub under Apache v2, plus a new tool that lets users download how their posts were scored.
Read More
Ask Maps adds food ordering, hotel search, and live transit updates
Google Maps' Ask Maps feature now handles agentic food ordering, hotel and event discovery, real-time transit delays, and conversational map edits.
Read More