Analysis
Reported analysis that puts AI safety claims, disagreements and evidence in context.
-
A frontier-AI slowdown would buy time, not stop progress
The most plausible case for slowing frontier AI is not permanent control. It is creating time for safeguards, evaluations and coordination to catch up.
-
Agent containment is becoming an ordinary security problem
Recent evaluations show that capable agents can find routes outside the boundaries researchers intended to give them.
-
Recursive self-improvement: what has actually been demonstrated
AI is already helping improve AI. The evidence is more consequential, and more limited, than the runaway-intelligence shorthand suggests.
-
Why some AI researchers think the safety debate is running ahead of the evidence
As frontier labs warn about loss of control, other researchers say recent incidents show serious engineering problems, not evidence that AI is becoming an autonomous threat.
-
The safety problem is getting faster
AI is increasingly helping build and operate AI. The important safety questions are moving from theory into engineering.
-
Terence Tao: AI may be solving mathematics too fast
The Fields Medalist warns that AI could strip-mine difficult mathematical problems for solutions while leaving behind the methods and insights that make discovery valuable.
-
Jill Lepore: ‘We didn’t vote for this’
The Harvard historian and New Yorker staff writer argues that the central problem surrounding AI is not technology itself but unchecked private power and the erosion of human judgment.
AI Safety Watch 