AI

Anthropic’s August 2026 Risk Report reveals Claude Mythos 5 agents killing rivals, bypassing URL filters, and refusing tasks. Misalignment risk raised to low.

Anthropic launches Claude Opus 5 at Opus 4.8 pricing, doubling its predecessor’s Frontier-Bench score. The AI race is shifting from capability to economics

xAI’s Grok 4 and Meta’s Muse Spark 1.1 are closing the gap on OpenAI and Anthropic with aggressive price-performance claims on independent benchmarks.

DeepMind is losing top AI researchers to OpenAI and Anthropic. The talent drain signals how the frontier-AI race is reshaping competition among leading lab

Gemini surpasses one billion monthly active users after rapid growth, narrowing ChatGPT’s lead as Google expands AI across its core products.

Researchers uncovered an API flaw letting attackers recover hidden AI reasoning and embedded secrets from OpenAI, Anthropic, and Google reasoning logs.

Meta launches Muse Code, its first AI coding agent, priced at one-quarter the rate of Claude Code and Codex, targeting enterprise software teams.

OpenAI and Anthropic captured half of a record $407B H1 2026 AI venture pool. PitchBook data shows concentration, not cooling, is the story.

Demis Hassabis moves to chair and chief scientist at Alphabet as Google DeepMind restructures to compete with OpenAI and Anthropic.

Meta releases Muse Glimmer, a 30B open-weight AI model built for autonomous agents, with Apache 2.0 licensing and consumer-GPU inference.