Regulatory prediction calibration
When Conrad spots an early regulatory signal — a consultation, a speech, a discussion paper — it records a dated, probabilistic prediction here before the outcome is known. We publish every prediction and score ourselves with the Brier score once it resolves: 0 is perfect, 0.25 is what you'd get from a coin flip at 50%, 1 is maximally wrong. We do not remove predictions that turn out wrong.
NIST will finalise or publish an updated Cyber AI Profile applying the Cybersecurity Framework 2.0 to AI-related cybersecurity risks.
Pending resolution
NIST CAISI will issue voluntary guidelines or best practices on security risks and mitigations for AI agent systems.
Pending resolution
ISO will develop or update AI standards or guidance to address consumer protection risks such as privacy, bias, autonomy, inclusion and misinformation safeguards.
Pending resolution
NIST will finalise its draft guidance on managing misuse risks for dual-use foundation models.
Pending resolution
NIST will revise the AI RMF Playbook after updating AI RMF 1.0
Pending resolution
NIST will finalise or update formal guidance on identifying and managing bias in artificial intelligence systems following stakeholder comments on draft Special Publication 1270.
Pending resolution
The EU will adopt or materially advance amendments to the AI Act simplifying implementation, adjusting high-risk AI timelines, clarifying AI literacy obligations, expanding sandboxes, and permitting safeguarded special-category data processing for bias detection and correction.
Pending resolution
NIST will publish a Critical Infrastructure profile or similar implementation guidance under the AI Risk Management Framework.
Pending resolution
OMB will issue implementation guidance for federal agency procurement of large language models under Executive Order 14319, including requirements tied to truth-seeking and ideological neutrality principles.
Pending resolution
NIST will publish further formal AI standards guidance or draft standards aligning the AI RMF with international AI governance and assurance standards
Pending resolution
NIST will finalise AI 800-1 guidance on managing deliberate misuse risks for dual-use foundation models, including CBRN, cyber, deception, CSAM and non-consensual intimate imagery risks.
Pending resolution
NIST will advance further formal AI standards or guidance work on priority areas including AI risk management, TEVV, synthetic content transparency, security, privacy, incident response, and training-data practices.
Pending resolution
The Australian Government will progress formal responsible AI regulatory settings or guidance to manage AI-related harms following the National AI Plan 2025.
Pending resolution
NIST will publish or coordinate further AI standards-related guidance on AI risk management, testing and evaluation, transparency, security, incident response, or training-data practices under the U.S. global engagement plan for AI standards.
Pending resolution
The Australian Government will consult on or introduce further formal AI safety or responsible AI regulatory settings for Australian organisations.
Pending resolution
The Australian Government will release further formal proposals or guidance on responsible AI adoption and harm mitigation as part of its National AI Plan implementation.
Pending resolution
NIST's U.S. AI Safety Institute will issue final or updated formal guidance on identifying, measuring, managing and governing misuse risks for dual-use foundation models, including CBRN, cyber, deception, CSAM and non-consensual imagery risks.
Pending resolution