Earlier this week, the company released a special version of Mythos with extra guardrails. It called this product Fable 5. According to Anthropic, the government claims to have found a “jailbreak,” or a way that users could bypass some of Fable 5’s guardrails.

Google and Amazon could cement their cloud-computing dominance as artificial intelligence gets more expensive and complex.

Avataar.ai has launched ‘Varya’, an indigenous video AI model developed in India, aimed at making advanced artificial intelligence more accessible and affordable across the country. Designed to cater to diverse industries and users,

Hippocratic AI, the generative AI company building the only model clinically safe and fast enough for patient-facing voice AI in clinical settings as benchmarked against major frontier models, today launched the LLM Model Training Residency.

Anthropic said it could not yet release the full Claude Mythos 5 model, so instead it created Fable 5 with baked in safeguards.

In roadshow materials released in the lead-up to the IPO, SpaceX bragged that it had released four major iterations of Grok, and talked up its popularity on X, the platform formerly known as Twitter.

Guardrails make the powerful model less useful for AI researchers, though the company said it would grant safeguard-free access to the science community.

Asked in which area of AI-related risk their bank was least prepared, 72% of bankers chose model kill-switch protocols (34%) or regulatory reporting of AI failures (38%). "Regulat

Most AI models are designed to be autoregressive—they generate text left to right one token at a time. DiffusionGemma has more in common with image generation models, which start with static and then denoise it to create the desired content.