Breaking

Jacob Coxon Warns New York Lawmakers on AI Control Risks as Council Weighs Safeguards

October 5, 2026 at 13:10 EDT

Policy & Regulation
Security
AI Agents

Former Anthropic researcher Jacob Coxon used a New York City Council hearing Monday to warn that leading AI labs are moving toward more autonomous, potentially self-improving systems without a proven way to keep them under human control. The testimony turned a debate inside frontier labs into a municipal policy fight over validation, shutdown capability and liability.

Oct. 5, 2026 · New York City Council

AI Control Risks Reach City Hall

A former frontier-lab researcher warned that autonomous systems are moving faster than proven safeguards. Lawmakers weighed validation, shutdown capability and liability rules.

council members convened

frontier firms under oath

AI-related bills on agenda

$25,000

per-instance penalty in one proposal

Whistleblower Incentive Scale

The proposed complaint system pays a share of recovered penalties, with a higher share when a complainant is designated to enforce.

25%

standard recovery

50%

designated enforcement

From Warning To Rules

Control warning

→

Under-oath hearing

→

Validation and shutdown

→

Reporting and liability

What supporters see

Independent validation before deployment
Human shutdown capability
24-hour incident reporting

What remains hard

Loss-of-control forecasts are disputed
Technical definitions are unsettled
City-level enforcement is untested

Continue reading

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