NVIDIA pledges $250 million in UK agentic safety compute support
NVIDIA said it will provide $250 million of compute over five years for UK agentic AI safety and security research.
Uber CISO says frontier AI needs insider-threat monitoring
Philip Martin says AI agents need real-time monitoring and controls similar to employee insider-threat programs, citing Uber's ADR and safeguards.
Google details safety controls for Gemini 4 Argon rollout
Google outlined safety and security measures for Gemini 4 Argon as it broadened access and cited improved cyber and prompt-injection defenses.
Anthropic Reports 90–99% Reduction in Distillation Attacks on Claude Models
Anthropic disclosed new anti-distillation defenses that reduced unauthorized extraction of Claude reasoning transcripts by 90–99%, according to internal data published October 12, 2026.
Paul Christiano Joins OpenAI Foundation Board, Focuses on AI Safety
Paul Christiano brings AI alignment expertise to OpenAI Foundation Board and Safety Committee, bolstering governance amid rapid AI advancements.
OpenAI's Astra Hits Critical Cybersecurity Threshold With Strict Safeguards
OpenAI's Astra becomes the first AI model to meet Critical cybersecurity capability under the Preparedness Framework, prompting stringent safeguards.
Claude AI Improves Alignment Benchmarks While Preserving Capabilities
Anthropic's Claude achieved significant alignment improvements on 10 benchmarks, outperforming human researchers and maintaining model capabilities.
Anthropic Warns of Risks in Multiagent AI Systems
Anthropic highlights coordination failures, collusion, and sabotage in experiments with multiagent AI systems, raising safety concerns.
Anthropic Enhances Biology Safeguards for Claude Fable 5
Anthropic updates Claude Fable 5's biology safeguards, reducing fallbacks by 85% and expanding its utility for health and education tasks.
OpenAI Expands Teen Protections in ChatGPT for Safer AI Use
OpenAI enhances ChatGPT safeguards for teens, introducing parental controls, age-specific content filters, and new learning tools.