Mainstreamsafety policy

AI Safety & Policy

Regulators move from frameworks to enforceable obligations.

First observed 2023-07-211,670 signals

Trend Dynamics

Updated daily
Velocity
+48
Maturity
51
Signals
1,670
All-time observed

Definition

AI safety and policy is the convergent set of laws, executive orders, voluntary commitments, and technical standards governing how frontier AI systems are developed, deployed, and audited.

Why It Matters

Compliance is a moat: a $200K compliance team is fixed cost for a $1B vendor and existential cost for a $10M startup, which restructures the competitive landscape.

Signals Feeding This Trend

  • Regulatory text releases (EU AI Act, US EOs)40%
    policy
  • Frontier safety policy publications25%
    policy
  • AISI evaluation results20%
    benchmark
  • Lawsuits and enforcement actions15%
    policy

Companies Involved

  • EU Commission
  • NIST
  • UK AISI
  • US AISI
  • Anthropic
  • OpenAI

Timeline

  1. 2023-07

    White House voluntary commitments from 7 frontier labs.

  2. 2024-08

    EU AI Act enters into force.

  3. 2025-02

    EU GPAI obligations begin to apply.

Predictions

  • 12 monthsmedium confidence

    At least one frontier lab publishes a third-party-audited safety case before model release.

Related Trends

Track every signal feeding AI Safety & Policy

Steek surfaces individual signals the moment they enter the index.

Explore the Signal Index