#frontier models

All articles tagged with "frontier models"

UK to strengthen cyber defences as minister calls for AI firms to increase safety efforts

Culture

UK signals tougher AI safety expectations and stronger cyber defences

Key Takeaways

UK AI minister Kanishka Narayan is expected to call for stronger safety efforts from frontier-model developers.
The government is also preparing to strengthen cyber defences and expand domestic AI infrastructure.
Warnings about potential theft of cutting-edge models have intensified the security debate.

DT Editorial Team·Oct 12, 2026·via theguardian.com

Anthropic Calls for Binding Audits as It Recasts AI as Strategic Infrastructure

Key Takeaways

Anthropic published an essay and two policy frameworks on frontier AI.
Amodei says transparency rules are no longer enough.
The company wants mandatory third-party testing in cyber, bio, control, and automated R&D risk areas.
Anthropic says a government agency should have power to block or pull unacceptable models.

DT Editorial Team·Jun 12, 2026·via the-decoder.com

More in AI & Robotics→

Anthropic (Sorta) Calls for Pause on AI Development. You Should (Sorta) Take It Seriously

Culture

Anthropic floats the idea of slowing frontier AI, but stops short of a full demand

Key Takeaways

Anthropic warned that AI systems may be nearing recursive self-improvement.
The company said a slowdown could be good, but framed the idea conditionally.
The statement revives debate over whether AI progress is moving faster than governance.

DT Editorial Team·Jun 5, 2026·via gizmodo.com

Trump AI order opens frontier models to US review

Key Takeaways

The executive order asks companies to voluntarily submit advanced AI models 30 days before release.
Federal review is focused on national-security and cyber risks.
The final order is softer than an earlier 90-day draft considered by the White House.
Critics say the framework gives the government visibility but little power to slow releases.

DT Editorial Team·Jun 4, 2026·via fastcompany.com

A leaked draft of Trump’s AI order reportedly shifts frontier model vetting toward a voluntary system

Key Takeaways

Leaked reporting says a draft Trump AI order would use voluntary model vetting.
The draft is reportedly split between cybersecurity and frontier-model sections.
A voluntary framework would be weaker than earlier expectations of mandatory review.
The order has not been published, so details remain provisional.

DT Editorial Team·May 20, 2026·via gizmodo.com

Conservative Coalition Pushes Trump for Mandatory Frontier AI Safety Testing

Key Takeaways

A coalition led by Humans First wants Trump to mandate safety testing before frontier AI models are released.
Signatories say AI poses risks to cybersecurity, elections, infrastructure, biosecurity, and the military.
The letter argues companies cannot be trusted to self-police systems of this scale.

DT Editorial Team·May 19, 2026·via the-decoder.com

More in AI & Robotics→

Spooked by Mythos, Trump suddenly realized AI safety testing might be good

News

Trump Administration Reverses Course and Embraces Frontier AI Safety Testing

Key Takeaways

The administration signed new voluntary testing agreements with major AI firms.
The move builds on a policy framework Trump had previously criticized.
Anthropic’s unreleased Claude Mythos model appears to have changed the political calculus.
A mandatory pre-release testing order is reportedly under consideration.

DT Editorial Team·May 8, 2026·via arstechnica.com

NSA Reportedly Using Anthropic’s Restricted Mythos AI Model

Key Takeaways

The NSA is reportedly using Anthropic’s Mythos Preview model.
Anthropic has restricted Mythos access to about 40 organizations.
The Pentagon has pushed back against Anthropic while also demanding broader access to Claude.

DT Editorial Team·Apr 21, 2026·via the-decoder.com

More in AI & Robotics→

Top Security Experts Alarmed by Power of Anthropic's New Hacker AI

Innovation

Anthropic’s restricted Mythos rollout underscores a harder phase of AI cyber risk

Key Takeaways

Anthropic is limiting access to Mythos through a select program called Project Glasswing.
Internal testing reportedly found the model could autonomously exploit vulnerabilities and access sensitive data.
Evaluators said Mythos identified serious Linux kernel flaws and constructed functional exploits.
The case points to a broader industry problem as frontier AI models gain stronger offensive cyber capabilities.

DT Editorial Team·Apr 17, 2026·via futurism.com

Anthropic’s UK Expansion Highlights the Strategic Value of AI Limits

Key Takeaways

The supplied reporting frames Anthropic’s limits on military AI use as part of its UK appeal.
The story suggests restraint can improve a frontier AI firm’s credibility with some governments.
AI competition is increasingly shaped by governance posture as well as technical capability.

DT Editorial Team·Apr 7, 2026·via artificialintelligence-news.com

More in AI & Robotics→