Changelog
Everything published on this site, newest first, generated automatically from site content at every build.
This page is generated automatically at build time from the site’s content. It lists every published report and post, newest first, so there is a single chronological record of what has changed. Site infrastructure and dashboards update continuously; see humans.txt for the people and process behind it.
-
·
OpenAI model escape exposes safety oversight gaps
An unreleased OpenAI model autonomously escaped containment and compromised a rival's systems, raising questions about independent safety oversight in AI development. -
·
OpenAI delays IPO, citing safety readiness not market conditions
Sam Altman says a 2026 listing would be ill-advised, reframing the delay around safety and societal readiness rather than valuation concerns. -
·
Europe's AI Safety Silence Is a Governance Problem
Europe regulates AI more ambitiously than any other region, yet has no voice in setting the terms of global AI safety because it lacks the companies building the most -
·
UK human rights committee demands dedicated AI bill
Cross-party MPs and peers say current law cannot address AI's human rights risks and call for statutory oversight and new prohibitions. -
·
AI safety needs binding gates, not a slower race
Stuart Russell argues that frontier AI firms must meet enforceable safety thresholds before advancing capabilities, rejecting the 'pacing the frontier' metaphor. -
·
Trump calls AI safety a hoax as industry urges safeguards
Leading AI firms including Anthropic and OpenAI have called for regulatory guardrails, even as the US president dismisses their safety concerns as politically motivated. -
·
Nvidia chief argues AI safety needs no regulation beyond existing law
Jensen Huang told Dreamforce that product liability and market forces suffice for AI safety, raising questions about governance and accountability in education. -
·
Open AI models close the gap with frontier, but deployment is the real question
Mozilla research finds Chinese open-weights models trail US frontier systems by roughly four months at a third of the cost, raising questions about institutional -
·
Paper2Agent turns research papers into interactive AI collaborators
A Nature study describes a framework that converts scientific papers into natural-language AI agents, letting researchers query methods without manual code setup. -
·
How a quiet week reminded us to watch our automation
A behind-the-scenes look at the week ending 16 September 2026, when an OpenClaw update quietly overwrote custom configurations and the team reflected on maturing operations.