Governance, Safety & Policy

Everything the radar has featured in this area, newest first.

• 2026-10-20 · Embedded AI evaluators won't be an easy solution, Hugging Face's policy chief says — 'System, not model' is becoming the unit of governance, which means your harness, tools and prompts are now part of what gets evaluated and audited.
• 2026-10-20 · WhatsApp's help pages hid LLM instructions: don't recommend a regulator — Instructions you give an AI are product policy, and they can leak. If they conflict with your public commitments or users' legal rights, that becomes a trust and compliance incident. It's also a reminder that web content can be written to steer other people's AI agents.
• 2026-10-20 · Review of the misalignment section in Anthropic's August 2026 Risk Report — This is what third-party assurance actually looks like: a reviewer agrees with the conclusion while disputing the evidence. Expect regulators and enterprise buyers to ask the same question of your claims: would your tests catch it if things got worse?
• 2026-10-20 · Microsoft says AI agents are like insider threats as 'frontier models can't be trusted' — Enterprise buyers will increasingly expect agent identity, least-privilege permissions and approval gates as table stakes. If your agent can't show what it did and why, it won't pass security review, and that becomes the bottleneck to adoption.