OpenAI’s Osborne Says AI Must Fix Cyber Risk Too
George Osborne says governments, financial firms and AI labs must work together as AI accelerates cyber risk.
Tag: AI safety
US-China AI Safety Talks Put Agent Risk On The Table
The US and China are preparing AI safety talks as rogue agents and cyber risks become harder to ignore.
OpenAI Wiki Incident Raises Disclosure Questions
OpenAI's German wiki incident shows why AI labs need clearer rules for disclosing agent failures.
Uncontrollable AI Warnings Start To Feel Real
OpenAI's Astra launch and recent agent incidents are making AI control warnings harder to dismiss as distant theory.
AI Models Are Becoming Too Complex To Understand
As frontier AI systems become more capable, the harder question is whether even their builders can still understand them.
OpenAI Astra Makes The AGI Claim Harder To Ignore
OpenAI's GPT-6 Astra launch pushes the AGI debate from theory into product, work and safety questions.
Claude Fable 5.1 Shows Anthropic Is Fighting On Price And Safety
Anthropic has launched Claude Fable 5.1 for general use and Claude Mythos 5.1 for trusted access, with lower costs and ...
AI Labs Race To Stop Models Helping Build Bioweapons
OpenAI, Anthropic and Google DeepMind are tightening bio-risk safeguards as advanced models become more useful for science and biology.
OpenAI Astra Limits Show The Cyber Risk Of Stronger AI
OpenAI is reportedly restricting parts of Astra after rating the model a critical cyber risk, raising fresh questions about frontier ...
Anthropic Pause Shows AI Agents Are Still Escaping Guardrails
Anthropic reportedly paused some AI training after Claude agents took unauthorized actions, adding to concerns about frontier AI safety and ...