🟡 Medium | Source: The Register — Security
OpenAI has committed to integrating Project Astra-style safety controls into its AI systems, while Anthropic has relaxed restrictions on its Fable AI model, allowing it to engage with a broader range of content scenarios. The moves reflect diverging approaches among leading AI labs to balancing capability with safety guardrails. For security teams, this signals a shifting threat landscape as AI models are granted greater autonomy and fewer behavioural constraints.
Security Architect’s Take: Review your organisation’s AI usage policies and assess whether models you consume via API — particularly Anthropic’s Claude underpinning Fable — have had their safety constraints updated; audit permitted use cases and ensure your AI governance framework accounts for models with loosened restrictions.
Original advisory: OpenAI pledges to add Astra security as Anthropic loosens Fable’s leash