OpenAI backs shared standards for advanced AI safety

Google DeepMind has published an AI Control Roadmap focused on securing internal systems that deploy AI agents, combining traditional safeguards with real-time monitoring approaches. The roadmap addresses the challenge of maintaining control over increasingly autonomous AI systems as they take on more complex tasks. This represents a shift toward proactive security frameworks designed to prevent misuse or unintended behavior in production AI agent deployments.

Google researchers propose 'faithful uncertainty,' a technique that allows large language models to express qualified guesses rather than either confidently hallucinating or refusing to answer. The approach reframes hallucinations as 'confident errors' and enables models to hedge responses appropriately, preserving utility while maintaining trustworthiness. This addresses a core tradeoff in LLM deployment where eliminating factual errors typically forces models to abstain from answering questions they actually know.

Google DeepMind and partners have announced a $10M funding call dedicated to multi-agent AI safety research. The initiative aims to address safety challenges that emerge when multiple AI systems interact with each other. This represents a targeted investment in a research area that has received less attention than single-agent safety concerns.

Anthropic CEO Dario Amodei published a policy essay calling for FAA-style government regulation of powerful AI models, arguing that mandatory third-party testing and potential deployment blocks are necessary as AI capabilities grow. The company released two policy frameworks alongside its most powerful model yet, Claude Fable 5, and a gated version with advanced cyber capabilities. The proposal would require models trained with over 10^25 FLOPs or developed by companies with $500 million in AI revenue to undergo mandatory testing, with regulators able to block or delay deployment if severe risks are identified.