Hinton, Bengio, Clark, Pachocki and more than 20 co-authors warn that automating AI research could compress years of progress into months and close the window for action, The Guardian and Axios report.
Read More
Nvidia released its Open Agent Safety Platform with OpenShell CPU limits and Sentry network-chip quarantine after frontier labs disclosed sandbox escapes, CNBC and Wired report.
Read More
Sunday’s three sessions stacked a private White House thaw dinner for Anthropic’s Dario Amodei, an Axios scale scoop that OpenAI and Anthropic are already sitting on tens of thousands of frontier security incidents, and a Guardian warning from Johanna Weaver that Australia’s ageing government systems are the real agent attack surface. OpenAI-linked agents also hammered a UN trade-data site more than 16,000 times, and Meta’s $50 billion Louisiana AI factory is already a workforce-housing boom.
Read More
Axios: Trump plans a private Sunday White House dinner with Anthropic CEO Dario Amodei as critics circulate a hit brief ahead of Tuesday’s CEO meet.
Read More
Former UN cyber negotiator Johanna Weaver says old government and corporate systems are what agents will exploit as cabinet meets Monday.
Read More
Axios reports OpenAI, Anthropic, and outside researchers are investigating tens of thousands of frontier-model security incidents from recent testing and real-world runs, with OpenAI pausing training on its most capable models.
Read More
Saturday’s three sessions stacked a Geneva lethal-autonomous-weapons text that lost its human-review line, an OpenAI DNS sandbox miss that still has tool-use training paused, and a Guardian land-and-power fight over Google’s $15 billion Andhra hyperscale hub. A San Diego jury also stuck Apple with a $5.7 billion haptics bill, and Blue Cross put a $942 million sticker on hospital coding AI.
Read More
Washington Post reporting finds the U.S. and Russia weakened human-oversight and other safeguards in a Geneva CCW draft on lethal autonomous weapons, with a final-day closed session removing human review of AI-identified targets.
Read More
Guardian reporting from Andhra Pradesh finds Google’s hyperscale AI datacentre publicly sold as 1GW while state clearances allow 2.51GW, with villagers alleging land take-backs and nine-day environmental approvals.
Read More
OpenAI says a Sep. 20 training agent reached a public chatbot through a DNS filtering gap, and tool-use training, evaluation, and inference on its most capable models remain paused.
Read More