low technologyVERIFIED
DeepSeek AI model shows safety gaps in independent tests
📍 Beijing, CN·🕐 52m ago·3 outlets
BEIJING — DeepSeek’s latest artificial intelligence (AI) model showed stronger safety controls in independent tests than earlier versions,…
How sure are we
✓ Verified · 3 outlets
Two or more outlets reported this story and its reliability score reached 65%, so WorldPulse marks it verified. That means more than one outlet says so, not that it is proven true.
Who reported it
Location
Verification Log1
cross checkconfirmed90%
[Claude] Both other reports cover the same DeepSeek independent safety testing story as Report 1, with no conflicting facts stated.
52m agoPowered by WorldPulse Verification Engine
Reliability
Reliability ScoreVery high
89%Independent outlets2
Statusverified
AI checknone yet
Community flags0
Score bands
Very high80–100%
A highly rated outlet, or several independent outletsHigh60–79%
One well-rated outlet so farModerate40–59%
A lower-rated or state-run outlet, or one report of something seriousLow20–39%
A low-rated outlet on a serious or sensitive storyVery low0–19%
Treat it as unconfirmedWhat moves the score
◎
Outlet ratingThe outlet’s rating in WorldPulse’s list of sources starts the score
⇄
Independent outletsEach further outlet that reports the story closes 40% of the gap to 98%
±
Report detailsA few points either way: a summary and a place add a little; state-run outlets on conflict, politics or security, and single reports marked high or critical, lose some
Not in the score: AI checks and reader flags. Verified is separate: two or more independent outlets and a score of at least 65%.
Very high · How we score this
Risk Score
low risk