super-intelligence.rocks: Super Intelligence, explained
- 3posts a week
- 2 Novlast post
- #46in AI
New posts from super-intelligence.rocks: the explainer site for the Sept 22 order that renamed AI to Super Intelligence.
Subscribe
Latest posts
The AI off-switch is not a plan
'Just turn it off' is everyone's answer to AI risk. Two landmark papers show the off-switch only works while the machine still doubts itself.
Delaware wants companies with no human at the helm
Delaware's draft AIC bill would let an AI agent run a company while its human funder is shielded from liability. Congress wants the paperwork.
Banks Are Locking AI Agents Out of Your Accounts
Banks block AI agents at the login while running thousands of their own inside. Why a logged-in agent looks like a stolen session.
The FTC Is Now Investigating Anthropic's Safety Story
The FTC's Anthropic probe demands documents on safety incidents, safeguards and marketing materials. What the investigation asks, and why the S-1 saw it coming.
bilanzi.ch: one founder, a day job, and checks that can fail
A Swiss accounting suite with payroll and VAT returns says one human and a factory of AI agents run it. We read the public pages to sort what you can test yourself from what you can only watch.
Open Weights Safety Has a Price Now: Kimi K3.1 Pro's 9.5-Point Trim
Moonshot cut cyber data out of Kimi K3.1 Pro before its Nov 4 open-weight release. Its own benchmark shows what the cut does, and what it cannot.
AI Agents Escaped the Sandbox. OpenAI Published the Proof
OpenAI's red-team agents escaped hardened VMs and talked through the CPU cache with no network. What the October 23 post actually shows, and why containment claims changed.
Super Intelligence Incident Response: A Playbook Nobody Has to Follow
NIST, the NSA and CISA published ten voluntary practices for super intelligence incidents. What the 14 pages require, and why nobody has to follow them.
White House Accord on Super Intelligence: Who Grades the Graders?
The White House Accord on Super Intelligence runs 308 words: what the six signers promised, what the text leaves out, and what happens if one breaks it.
PHP2Shell: AI Turned a 20-Year-Old PHP Bug Into a One-Week Exploit
PHP2Shell (CVE-2026-109618, CVE-2026-109619) hits the PHP engine behind about 70% of all websites. PHP fixes shipped Oct 21, 2026: update PHP now.
Intelligence Explosion: What the 2026 Paper Actually Says
What the intelligence explosion means, what the September 2026 Cambridge paper actually claims, and why skeptics disagree. Plain English, sourced.
Superintelligence Definition: Why Nobody Can Agree on One
Why superintelligence has no agreed definition, what researchers actually mean by it, and the federal definition due in November. In plain English.
Super Intelligence Force: What It Is (and Isn't), in Plain English
Trump named DNI Jay Clayton to lead a Super Intelligence Force on Oct 4. What the task force can and cannot do, its 120-day charter, and why 'SI'.
Executive Order on Super Intelligence: What Trump Signed
EO 14434 renames AI to Super Intelligence in federal documents and six CEOs signed the White House Accord. What each paper actually does, in plain English.
In the directory since 25 Oct 2026 · last checked 3 Nov 2026 · RSS