THE THINKING BEHIND THE SIGNAL

A strong signal needs
a clear explanation.

METHODOLOGY IN DEVELOPMENT

Two separate lenses.
Cyber conditions and AI capability.

Cyber Conditions describes evidenced disruption and escalation. The AI Capability Frontier describes breadth and reliability of ability. Neither is an official classification, a probability of a power cut, or a countdown. Higher AI capability does not automatically raise the cyber condition.

01

Observe the evidence

Prioritise confirmed essential-service incidents, credible cyber campaigns, actively exploited vulnerabilities and authoritative advisories.

02

Put it in context

Assess impact, reach, relevance and recency. Deduplicate reporting. Keep source confidence separate from threat severity.

03

Explain every movement

Publish the factors behind a change, when sources were checked and what is uncertain. Missing data must never silently become an “all clear”.

THIS PREVIEW

Feeds and assessments have different roles.

Both current levels are editorial assessments set manually by the site owner, with their dates and notes available in the scale information panels. Cyber Conditions covers the global picture; it is not an automated feed score. Service feeds update independently: an outage, a vulnerability warning and a confirmed cyber incident are different forms of evidence.

Cyber assessments must name their geographic scope. They can move up or down and can skip stages when the evidence warrants it. Missing data means “assessment unavailable”.

CYBER CONDITIONS

A SEPARATE CAPABILITY FRAMEWORK

Understanding the AI frontier.

Our proposed six stages describe capability milestones, including recursive self-improvement (RSI). They do not measure consciousness, intention or safety. A model may excel in one task while remaining limited elsewhere.

“Broad Capacities” sits immediately below RSI, followed by AGI and Superintelligent. RSI describes improvement of capabilities and of the process for further improvement. This sequence is our editorial framework; it is not an established scientific progression. Generality and performance still need separate evidence.

Read the research on levels of AGI ↗

FOLLOW THE ORIGINAL EVIDENCE

Sources worth knowing.

Country shading summarises disruption at monitored Cloudflare locations and active Google Cloud regions, not the availability of every service in a country. Missing or stale source coverage is shown as unknown. Red dots represent dated reviewed cases and automatically located news or official briefing reports. Automated reports stay at the smallest impact size and do not imply an ongoing attack; nationwide advisories have no invented location. Independent project. No affiliation with the organisations above. Assessments are kept separate from commercial product recommendations. Household guidance is currently UK-specific; international editions are planned. Contains NCSC information licensed under the Open Government Licence v3.0.