Breaking

Trump and AI Leaders Sign Voluntary 'Super Intelligence' Safety Accord

September 30, 2026 at 05:47 EDT

Policy & Regulation
Foundation Models
AI Agents

President Donald Trump and the heads of six major AI companies signed a voluntary White House Accord on Super Intelligence on Sept. 29, making company-led oversight the administration's immediate answer to rising concern over frontier AI safety. The pact commits OpenAI, Anthropic, Google, Meta, xAI and Nvidia to four layers of controls and audits, but leaves enforcement outside a binding regulatory regime.

September 29, 2026 · U.S. AI policy

A voluntary safety pact for frontier AI

Trump and six AI companies put four layers of company-led oversight at the center of frontier-model safety, even as agents become more autonomous and safety incidents keep pressure on the industry.

companies signed

control layers

people discussed for oversight

3-4

days to name AI czar

Cyber jailbreak refusal rates

Astra-class safeguards were described as refusing more test requests than GPT-5.6 Sol.

59%

GPT-5.6 Sol

91.5%

Astra

The oversight debate is moving from policy principles to measurable model behavior.

Blocks approximate ten percentage points each.

Four-layer accord structure

Internal controls

Monitor model behavior and threats.

→

Internal team

Check detection and remediation.

→

External auditor

Independently test controls.

→

Board committee

Receive reports and oversee fixes.

Why supporters see progress

Common baseline for frontier labs
Outside evaluation enters the pact
Board-level attention becomes explicit

Why critics are wary

Commitments are voluntary
Audit disclosure is not specified
Mandatory incident reporting remains unresolved

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.