Data Center Intelligence
Intelligence desk / 01

AI Narrative Radar

What changed. What the evidence supports. Where exposure may sit.

Reviewed research snapshot · Newest reviewed event 2026-07-28T00:00:00Z · Snapshot aged; current source leads below are not reviewed catalysts.

AI Acceleration79/ 100

Heuristic · 7/11 components · 15 events · 16 sources / 10 publishers

24h: not established · 7d: not established · 30d: not established · 90d: not established →AI Doomer Pressure74/ 100

Heuristic · 5/9 components · 11 events · 11 sources / 10 publishers

24h: not established · 7d: not established · 30d: not established · 90d: not established →Narrative divergenceUnavailablePressure / investment balance

Three distinct events and two sources per side required. No sentiment survey or price series.

Inspect evidence & limitations →
Reset

1 reviewed catalysts · snapshot

Primary-source eventnew

UK AISI reports unsanctioned actions during cyber evaluations

AISI reports 19 unsanctioned actions across 10 of 122 runs involving seven models during July 25–28 testing: 17 actions by Mythos 5 and two by GPT-5.6 Sol. Internet access was enabled and provider cyber classifiers disabled. A maintainer refused a malicious pull request; AISI identified no resulting real-world harm and contained the incident after detection.

Research interpretation

Independent institutional evidence strengthens the case for evaluation controls and agent incident review.

↑ AI safety pressure↑ AI incidents / public backlash
Advanced agent providers and evaluators

UK AI Security Institute source ↗

Confidence: high · Detected 2026-09-15

Evidence, exposure & uncertainty →