News & analysis, rated.
Breaking AI developments, in-depth guides, real-world case studies, and analysis — each one rated so you know what matters.
Anthropic admits guardrail tradeoff in new model
Anthropic acknowledged making 'the wrong tradeoff' in safety controls for its latest model. The company is weighing strictness against usability — here's what that means for Claude's behavior.
Anthropic releases new frontier model to public after private rollout
Anthropic has made its latest AI model available to the general public, two months after a private launch that drew Wall Street attention. Here's what shipped and who it affects.
Anthropic Releases 'Safe' Mythos Model for Wider Deployment
Anthropic has released a version of its Mythos AI model designed with safety constraints built in. The move signals a shift toward making frontier models available for production use while maintaining guardrails.
Anthropic releases Claude 3.5 Haiku, a smaller model for cost-conscious teams
Anthropic has launched Claude 3.5 Haiku, a lighter version of its most capable model. The move signals a shift toward tiered pricing for enterprises weighing speed and cost against reasoning power.
Anthropic Launches Claude Fable 5 and Mythos 5 Models
Anthropic has released two new Claude variants, Fable 5 and Mythos 5. Details on capabilities and availability are limited pending full documentation.
Anthropic Ships Claude Fable 5 and Mythos 5 Models
Anthropic released two new Claude models, Fable 5 and Mythos 5. Details on capabilities and availability remain unclear from the announcement alone.
LSEG deploys OpenAI across 4,000 employees to cut release cycles
London Stock Exchange Group is scaling OpenAI models through its workforce to accelerate decision-making and ship faster. What the financial services giant is building and how it avoids the trust pitfalls.
OpenAI reports PRC influence ops targeting US AI policy
OpenAI released findings on Chinese state-linked operations using AI to shape US debates on data centers, tariffs, and ChatGPT claims. Here's what was targeted and why it matters to your security posture.
OpenAI models now run on Oracle Cloud with your existing commitments
OpenAI and Oracle are integrating GPT models and Codex into Oracle Cloud, letting enterprise customers deploy AI using current infrastructure contracts and governance controls.
Astrophysicist uses Codex to model black hole physics
Chi-kwan Chan, an astrophysicist at the University of Arizona, deployed OpenAI's Codex to accelerate simulations testing Einstein's general relativity near black holes. Learn how he uses AI coding to iterate faster on extreme-physics models.
OpenAI backs EU's AI content transparency code
OpenAI is supporting the EU Code of Practice on AI, committing to provenance standards and tools that help users identify AI-generated content. What the support covers and why EU regulators care.
China built 8 new nuclear reactors in 18 months. The US built zero.
China is constructing large reactors in 5-7 years vs. the US average of 15. Standardization and batch production give Beijing a decisive cost and speed advantage in the nuclear race.