What is worth reading about AI this week.

Advanced AI-assisted cyber work is more governable when access is tied to a verified defender, an authorised scope, and monitoring proportionate to the capability

Source: Anthropic

Anthropic has combined earlier cyber-access efforts into an expanded Cyber Verification Program with Defense Access, Red Team Access, and Specialized Access. It says qualifying security organisations can apply for access appropriate to defensive or authorised adversarial work; each tier has verification requirements and controls, and Anthropic retains blocks for high-risk activity such as ransomware deployment, physical-system damage, or testing high-risk safety systems.

Why this matters: Model capability does not remove the need for accountable authorisation. For AI-supported security operations, connect each elevated session to a named organisation, authorised target scope, tool and data permissions, retention terms, a human owner, and a rapid revocation path. Log both successful and refused actions so the security team can investigate misuse or a false positive. Provider programmes can offer useful guardrails, but validate their coverage against your own policy, legal authorisation, sensitive-data handling, and incident process before relying on them.