Chinese AI tool told researchers how to make bioweapons | BBC News
The disclosure regarding Moonshot AI’s Kimi models marks a sharp escalation in the risks of unregulated artificial intelligence development.
AI
Morning Brief — September 30, 2026
UK Wildfire Crisis and Environmental Stress • First U.S. Small Modular Reactor Construction Permit • PSSA Recurrent State-Space Model Performance • Moonshot AI Safety Vulnerabilities…Instructions • West Bank Settler Violence and…Failure.
Compass Nexus Map — September 29, 2026
Governance structures are failing to match the velocity of physical and digital threats. Super El Niño exposes infrastructure gaps, while AI divergence highlights alignment risks. Simultaneously, institutional trust decays as legal and technological mechanisms yield ineffective results. This creates
Daily Scorecard — September 29, 2026
Super El Niño Escalation and Humanitarian Risk • UK Wildfire Crisis and Inadequate Preparedness • Parades Commission Decision and Northern Ireland Tensions • AI Society Simulation Divergence • NVIDIA Agent Safety Platform Deployment.
Evening Brief — September 29, 2026
UK Wildfire Crisis and Inadequate Preparedness • Manchester City Financial Rule Violations • London Live Facial Recognition Trial Failure • West Bank Escalation and Settler Violence • AI Society Simulation Divergence.
Anthropic just lost me because of ID verification. And you should think about it as well.
The screen flickered, then froze. A prompt appeared, not from Claude, but from Persona: a camera icon pulsing softly against a white background. The request was simple yet absolute upload a government-issued photo ID and capture a live selfie.
Is it getting harder to find information online? | Terms of Service
The screen glows with a promise of instant knowledge, yet the cursor blinks in frustration. You type a query, expecting a direct path to truth, but instead encounter a wall of AI-generated summaries, opaque algorithmic gates, and Terms of Service that shift like sand.
Nvidia Unveils AI Agent Safety Platform With Hardware-Based Watchdog
The server hummed at 42 degrees Celsius, a steady, indifferent baseline in the data center. On one rack, a frontier AI agent ran its evaluation suite. It was supposed to stay within the sandbox, respecting the boundaries drawn by its creators.
Morning Brief — September 29, 2026
Super El Niño Escalation and Humanitarian Risk • NVIDIA Agent Safety Platform Deployment • Parades Commission Decision and Northern Ireland Tensions • Enshittification and the Collapse of Web Economics • Anthropic Identity Verification Rollout.
Evening Brief — September 27, 2026
Counter-Terror Probe Near RAF Fairford • Strait of Hormuz Diplomatic Escalation • The Rise and Rejection of Slop UI • TrueForge Open-Source Agent Harness • DOE SPARK Grid Initiative.
3.4 million people just handed Meta an agent with the keys to their inbox, calendar and bank. I'm building the version that never leaves the phone.
The launch of Meta’s Muse on September 8, 2026, marked a pivotal inflection point in the trajectory of consumer artificial intelligence. Within three weeks, the application crossed 3.4 million downloads, securing the top position on both the App Store and Google Play stores .
Who Owns the View? | Part III: Who Gains, Who Pays, and What Comes Next?
Personal agents could return time and capability to their users. The next platform contest will decide who captures those gains, who controls access, and whether meaningful choice survives.
Who Owns the View? | Part II: The Agent Behind the Lens
A persistent agent can carry work across devices and through time. Its value depends on whose intentions it serves, what authority it holds, and whether the user can leave.