Every marker is a published Human-Level signal with its sources, timestamps, and how precisely its location is known. The Observatory shows what has been reviewed and published, not everything that is happening.
Stale data
Data as of
11 Nov 2026, 12:47 UTC
Source freshness
Stale data
Last source check
11 Nov 2026, 12:15 UTC
Update check
Checking…
Updates are checked periodically; new signals appear within minutes of being published, after editorial review or an automated check. Live streaming is not configured on this host.
Coverage: 44 signals: 42 reviewed by editors, 2 published automatically after an AI check. 5 sources are monitored for new items; each is checked before it is published. This is a curated selection, not a complete record of global events.
Loading map…
Legend
AISignal at a known city or public site
AIApproximate location (region or country) — not an exact site
3Several signals nearby; select to zoom in
AI AI Risk
Bi Biosecurity
Cy Cyber Warfare
Co Cognitive Manipulation
Su Surveillance State
En Energy Systems
Ev Environment
Tr Transhumanism
Colour and two-letter codes show the domain. Markers do not encode severity or confidence; there is no published scoring method for either.
AI Risk
Training activation classifiers for safety and alignment
On 2 Nov 2026 OpenAI's Alignment Research Blog published "Training activation classifiers for safety and alignment" by Alex Makelov, Xiaojun Xu, Sundaram Ananthanarayanan, Ethan Mcdonald and Tom Dupre la Tour. The post describes probes, small classifiers that read a monitored model's internal activations and can interrupt responses and block tool calls immediately. It says the post introduces the probes as efficient real-time monitors, and says the probes complement other safeguards rather than replace them. It also says they detect only what they are trained to detect and need retraining for new base models and as model weights change. The post says activation classifiers were added to the safety stack starting from GPT-5.6. The prose gives no recall or escalation numbers; the only numeric result is a footnote attributing most of the spread in probe performance to initialisation randomness (about 85% variance share). The authors say the figures are not necessarily representative of deployed classifiers. Company-reported; no independent evaluation found.
Source-reported fact
Primary source
Location
Not established · Location unknown
Event time
Unknown
Source published
Unknown
Last updated
11 Nov 2026, 06:07 UTC · revision 3
Loading evidence…
All matching signals (44)
Mapped signals (9)
Location not established (not shown on the map) (35)
How this map works
What appears
Signals an editor has reviewed and published, and items from monitored official sources that passed an automated AI check, which are labelled “Published automatically”. Each one cites at least one public source, or is labelled as editorial analysis with its methodology. Withdrawn signals are removed.
Locations
We use the least precise location that is still useful. Dashed markers are approximate (region or country) and do not mark an exact site. Signals without an established location are listed but not pinned. We do not map personal locations.
Timestamps
Event time, source publication time, and the time we retrieved a source are shown separately. Missing values are shown as unknown, never guessed.
Freshness and updates
The page checks for new data every minute. Source freshness is reported on its own: a working update check does not mean the sources are current. New signals appear after review and a site release.
No scores
Markers do not encode severity or confidence. Evidence is described qualitatively (for example “single source” or “multiple independent sources”) with a written rationale.
Domains
AI Risk: Capability releases, alignment failures, and loss-of-control concerns.
Biosecurity: Synthetic biology, pathogens, and dual-use life-science developments.
Cyber Warfare: Offensive AI, autonomous weapons, and attacks on critical infrastructure.
Cognitive Manipulation: Information operations, synthetic media, and psychological influence.
Surveillance State: State and corporate surveillance infrastructure and governance.
Energy Systems: Grid fragility, compute energy demand, and resource competition.
Environment: Climate tipping points, geoengineering, and ecological systems.
Transhumanism: Human augmentation, neurotechnology, and brain-computer interfaces.
Basemap: Natural Earth 1:110m countries (public domain).
