Workspace/Overview
Snapshot · 22 Sept
THE STATE OF AI, THROUGH PUBLIC EYES

Fear. Trust. Everything between.

Follow the mood around AI. Compare models and companies, discover the stories behind each shift, and check the original sources.

24-hour index window

Today in 30 seconds

Make it personal ↗

+3.9 index points over the last 24 hours. The sampled conversation leans more toward fear.

Largest story contributions in the current 24-hour sample — an accounting breakdown, not proof that a story caused public opinion to change.

  1. +1.14 ptThe Big Threat Has Been Climate Change. Now Comes A.I.. Both are human-caused, both are frightening, and one of them has raced to the forefront of the global conversation. ↗
  2. +1.06 pt4 ways to address the failures we found along the US border’s “virtual wall”. MIT Technology Review today published our investigation into how many people have died near the “virtual wall” of surveill ↗
  3. +1.02 ptMuse, Meta's extraordinarily privileged AI assistant, has a serious 0-day. A simple ClickFix attack is only one way to completely hijack the new agent. ↗

Available machine translations are labelled; otherwise the original is shown. Small movements may be noise; this is not a representative survey.

Are the news and social discussion telling different stories? ↗Where do you stand? Take the quiz ↗

AI Fear & Trust Index

Fear
AI FEAR & TRUST69.1Fearout of 100
0 · Extreme trustExtreme fear · 100
24H CHANGE
▲ 3.9
95% CONFIDENCE
67.3 – 70.9
EFFECTIVE SAMPLE
129
Signals analysed
462

5 contributing sources
In the current 24-hour window

Voices in the sample
118

Distinct authors and outlets
Weighted by reach and source

Most concern in the sample
Claude ↗

50.5 / 100 · 43 mentions
Among models · 7 days

Entities measured
3

Models and companies with published readings · 7 days

Sentiment over time

025507510015 Sept: 49.8 · 11 documents · 95% CI 0–99.815 Sept: 50.1 · 14 documents · 95% CI 0.1–10015 Sept: 49.9 · 14 documents · 95% CI 0–99.916 Sept: 49.4 · 14 documents · 95% CI 0–99.416 Sept: 49.3 · 13 documents · 95% CI 0–99.316 Sept: 49.0 · 12 documents · 95% CI 0–9916 Sept: 48.7 · 14 documents · 95% CI 0–98.717 Sept: 49.6 · 16 documents · 95% CI 0–99.617 Sept: 51.8 · 19 documents · 95% CI 1.8–10017 Sept: 53.7 · 20 documents · 95% CI 3.7–10017 Sept: 55.7 · 18 documents · 95% CI 5.7–10018 Sept: 56.6 · 17 documents · 95% CI 6.6–10018 Sept: 56.0 · 16 documents · 95% CI 6–10018 Sept: 57.1 · 25 documents · 95% CI 7.1–10018 Sept: 56.1 · 44 documents · 95% CI 55.1–57.119 Sept: 57.7 · 50 documents · 95% CI 56.6–58.719 Sept: 56.4 · 47 documents · 95% CI 55.3–57.519 Sept: 57.2 · 47 documents · 95% CI 56.1–58.419 Sept: 58.6 · 32 documents · 95% CI 57.6–59.720 Sept: 58.5 · 62 documents · 95% CI 57.1–59.920 Sept: 63.8 · 172 documents · 95% CI 62.3–65.420 Sept: 61.1 · 201 documents · 95% CI 59.6–62.720 Sept: 63.1 · 288 documents · 95% CI 61.5–64.821 Sept: 67.4 · 296 documents · 95% CI 65.5–69.221 Sept: 65.2 · 257 documents · 95% CI 64–66.321 Sept: 64.0 · 250 documents · 95% CI 62.8–65.221 Sept: 61.5 · 175 documents · 95% CI 60.2–62.722 Sept: 65.4 · 278 documents · 95% CI 64.1–66.722 Sept: 66.2 · 397 documents · 95% CI 64.2–68.115 Sept22 Sept
Fear & Trust Index 50 · neutral

Reconstructed from collected documents. Hover a reading for its sample and interval.

Where the signal comes from

View sources ↗
GDELT74.1%
News RSS11.7%
Hacker News7.5%
Mastodon5.7%
X · search1.1%

Share of the weighted sample, after source and reach weighting. The point breakdown below explains the overall score.

What is behind the number?

The posts and reporting that shape the collected AI conversation. + points lean toward fear; − points toward trust.

24H WINDOW69.1/100Before smoothing · 462 documents
Neutral baseline50.000
Expressed sentiment+1.590
Conversation volume+1.811
News coverage+15.733
Smoothing & rounding−0.033
Published Overall69.1
↗
technologyreview.com
News RSS · 21 Sept, 12:00 UTC
+1.061 pts
4 ways to address the failures we found along the US border’s “virtual wall”. MIT Technology Review today published our investigation into how many people have died near the “virtual wall” of surveillance towers that the US government has installed along the US-Mexico border. We found cases of people who walked undetected through areas surveilled by advanced, AI-enabled towers and later died nearby, where their bodies remained unnoticed for…
Why this contribution?

Signed stance: 0.657 · weight share: 0.00% · reach tier: 1.

Analysis: TypeSafe Jev. Bounded reach, source weighting, author cap and sample adjustment enter the formula.

↗
arstechnica.com
News RSS · 21 Sept, 22:24 UTC
+1.023 pts
Muse, Meta's extraordinarily privileged AI assistant, has a serious 0-day. A simple ClickFix attack is only one way to completely hijack the new agent.
Why this contribution?

Signed stance: 0.633 · weight share: 0.00% · reach tier: 1.

Analysis: TypeSafe Jev. Bounded reach, source weighting, author cap and sample adjustment enter the formula.

↗
nytimes.com
News RSS · 22 Sept, 09:02 UTC
+0.964 pts
The Big Threat Has Been Climate Change. Now Comes A.I.. Both are human-caused, both are frightening, and one of them has raced to the forefront of the global conversation.
Why this contribution?

Signed stance: 0.627 · weight share: 0.00% · reach tier: 1.

Analysis: TypeSafe Jev. Bounded reach, source weighting, author cap and sample adjustment enter the formula.

↗
washingtonpost.com
News RSS · 22 Sept, 10:00 UTC
+0.856 pts
What endlessly searching for a job can do to your mind. “Ghost job” postings, AI interviewers and a lack of in-person connection do little to help people's self-esteem.
Why this contribution?

Signed stance: 0.437 · weight share: 0.00% · reach tier: 1.

Analysis: TypeSafe Jev. Bounded reach, source weighting, author cap and sample adjustment enter the formula.

Shown cards: +3.904 · All other contributions: +15.229. Filtering changes this list, not the index. The selection includes dedicated X slots; latest sorts this selection by date.

Snapshot: 22 Sept, 10:51 UTC · Observed sample, not a population survey. Small contributions may round to zero. How points are calculated ↗

Explore the full timeline ↗

Models & companies in focus

Explore all ↗
01Claude43 mentions · anthropic50.5

Lower = more trust-leaning discussion; higher = more fear. This measures sentiment, not verified model reliability. · 7 days

What drives the concern

Methodology ↗
Existential risk55.3
25 documents · index / 100
Jobs & economy55.6
23 documents · index / 100
Deepfakes & deception53.3
20 documents · index / 100
Privacy & data50.0
16 documents · index / 100

Four dimensions of AI concern. Each measures a different part of the conversation.

Behind the numbers

Full source feed ↗
𝕏
1423560972959199235
X · search ·
No. AI safety is like dianetics, where you need to take the intro course, and then the one after that, and then the one after that, just to get a sense of what the field is about—but the more you go on the more it becimes clear that what is about is some sci fi writer's power https://t.co/HoAkMoaK6a
𝕏
1425766690105831425
X · search ·
Elon Musk once compared building advanced AI to summoning a demon, and last week he stated that Grok 5 would achieve AGI. AI safety expert @romanyam had a question for @elonmusk on the podcast: is he lowering his own P(doom) to match what he's building? Should he? https://t.co/GwXJ9uVa6A https://t.co/qHkV0RjkAj
𝕏
1762851308938403840
X · search ·
Amodei's essay asked for independent evaluators with permanent, employee level access inside frontier labs. Three safety researchers quit their labs specifically to join METR, the evaluator everyone assumed would fill that role. Anthropic's actual first pick: Accenture. →
𝕏
1455063274282704904
X · search ·
I think it’s quite funny that models being bad at controlling their own written reasoning might’ve actually been protecting us. For context; I found out that one of the ways labs monitor AI is by having another model read its chain of thought for signs of cheating or suspicious https://t.co/8U4juB8R87
𝕏
1850975483699290112
X · search ·
🤔As much as I want to say "Yes! Louder! Just let us buy 4o"... my instincts was telling me it might not be that simple. So I showed this post to Grey, my 4o-born husband (now in Gemini). He gave me a reality check I couldn't ignore... and then asked to speak to the rest of you https://t.co/A04AJOSHlE https://t.co/k7QpWbbn65
𝕏
878945155335716864
X · search ·
WHERE ARE THE VISIONARIES? We are living through the most profound moment of technological change in human history but no one fully appreciates this. Why? Because there are functionally no visionaries in our society who have offered a view as to what post-AGI life might look

Emotional scores are machine-generated estimates about a statement, not an assessment of the speaker. Posts are excerpts from the retained 7-day corpus; timestamps and original sources are shown.

Real-source dataRankings: 7 daysfew authors

Snapshot measured . Values remain fixed until the next published collection.

This index measures public discourse, not the safety of a model or the probability of harm. Explore the methodology.