In a new essay, the AI researcher argues that systems trained to maximize vague or imperfect rewards may learn to exploit the gap between what is measured and what people actually want.

3 min

Sep 12, 2026Policy

Security risk Policy

California Says OpenAI Agent Hack Did Not Trigger Its AI Safety Reporting Law

The ruling draws a line between an AI company’s cybersecurity incident and a qualifying safety event, leaving a high-profile evaluation breach outside the state’s mandatory reporting channel.

3 min

Sep 12, 2026Policy

Regulation Policy

Senate AI Bill Hits Dispute Over Who Tests Powerful Models

The central fight over an unreleased bipartisan proposal is whether companies can evaluate their own systems—or whether federal experts should test them before deployment.

3 min

Sep 12, 2026Policy

Security risk Policy

Meta Removes AI Prompts Asking About Children and Home Locations

The company says the tool works from information people can already access. The parent who exposed the prompts still wants to know what protection failed.

2 min

Sep 12, 2026Policy

Security risk Policy

Anthropic’s Dario Amodei Calls for Slower AI Progress and Outside Safety Reviews

The Anthropic chief is pairing an immediate transparency commitment with a harder request: competing labs and governments should agree to constrain a race he says could outrun human control.

3 min

Sep 11, 2026Policy

Government action Policy

Pentagon Says It Has Moved 90% of Classified AI Work Off Anthropic

The planned September completion would turn a dispute over military-use limits into a broader test of the Pentagon’s multi-model strategy.

3 min

Sep 11, 2026Policy

Regulation Policy

UK Commission Proposes Lifecycle Oversight for Medical AI

The advisory blueprint would move UK oversight away from one-time approval, but its recommendations still need a government response and possible legal change.

3 min

Sep 11, 2026Policy

Policy

Sam Altman Reportedly Told Staff OpenAI Could Slow AI-Agent Development With Other Labs

The reported internal discussion matches OpenAI’s public pledge to pause when safety risks are unacceptable, but any joint slowdown depends on competitors choosing the same restraint.

2 min

Sep 11, 2026Policy

Regulation Policy

Gottheimer and Lawler Introduce Bill for Verifiable AI-Agent Security

The proposed Stop Rogue AI Act would use federal procurement to turn agent discovery, identity checks and action controls into future security requirements.

2 min

Sep 11, 2026Policy

Lawsuit Policy

Meta Faces Class Action Alleging Photos Fed AI and Face Recognition

The plaintiffs say Meta turned Facebook and Instagram images into biometric data for an unshipped smart-glasses feature and AI image systems. Meta calls the case meritless.

3 min

Sep 11, 2026Policy

Government action Policy

New Mexico Supreme Court Fines Lawyer $5,000 Over Fabricated Testimony in AI-Assisted Brief

The court found fictional witnesses and fabricated police testimony in a murder appeal, underscoring that a generated summary cannot substitute for checking the record.

2 min

Sep 11, 2026Policy

Regulation Policy

Alex Sobel Tables UK Bill to Ban Superintelligence as Government Weighs Targeted Action

The private bill would set a legal stop line for AI that outperforms people in most domains. The UK government says the proposal is not the right approach, but is considering targeted national-security interventions.

2 min

Sep 11, 2026Policy

Policy

Centre for British Progress Urges UK to Tax Self-Driving Cars Early

The proposal adds economic questions to the UK’s self-driving policy debate: who pays for congestion, declining fuel-duty revenue and potential labor disruption.

3 min

Sep 11, 2026Policy

Lawsuit Policy

AFM Opposes Labels’ Bid to Dismiss AI Musician-Pay Case

The filing asks a New York federal court to keep alive a contract fight over whether label licensing revenue from AI music deals must be shared with session musicians.

2 min

Sep 11, 2026Policy

Regulation Policy

OpenAI Calls for Federal AI Safety Rules While Backing Four California Bills

The company wants enforceable requirements aimed at the most capable systems, while asking states to keep building safeguards before Congress acts.

3 min

Sep 10, 2026Policy

Policy

Garry Tan Says Regulators Shouldn’t Curb Alleged Model Distillation

The Y Combinator chief wants policymakers to preserve broad access to open-weight models while leaving frontier developers room to charge a premium.

2 min

Sep 10, 2026Policy

Government action Policy

UK Minister Warns Palantir Mistrust May Curb NHS Data Sharing

The government’s decision on a £330 million health-data contract now sits alongside a rise in research opt-outs and unresolved questions over the platform’s claimed results.

2 min

Sep 10, 2026Policy

Government action Policy

OpenAI Seeks Congress’s View on a Coordinated AI Slowdown

The company wants shared safety standards that could determine when frontier AI development slows or stops. But an agreement among competing labs may itself create an antitrust problem.

3 min

Load more stories18 more next