0:00
New York City just got called out again
0:02
for how it handles artificial
0:04
intelligence. State Controller Tom Da
0:07
Napoli's office first flagged this back
0:09
in 2023, warning that the city's
0:12
approach to AI was incomplete and ad
0:14
hoc. 3 years later, a new follow-up
0:17
audit finds the city has only partially
0:19
fixed it. To be fair, there's been some
0:22
progress. The city built out a working
0:24
definition of what counts as AI,
0:27
published early guidance on generative
0:29
AI tools, and formed a city-wide
0:31
steering committee. But the core
0:33
problems remain. There's still no
0:35
complete inventory of which AI tools its
0:38
agencies actually use. There's still no
0:41
consistent city-wide standard for how
0:43
those tools get approved. And there's
0:45
still no formal way for the public to
0:47
file a complaint if an AI system gets
0:49
something wrong about them. Here's why
0:52
that matters. The agencies in question
0:54
aren't minor ones. We're talking the
0:56
NYPD, the Department of Education, the
0:59
Department of Buildings and
1:00
Administration for Children's Services,
1:03
agencies making decisions about
1:04
policing, our kids schools, building
1:06
safety, and child welfare. And right
1:09
now, much of that oversight runs on the
1:11
honor system. Agencies are largely
1:13
trusted to self-report which AI tools
1:16
they're even using. The city's tech
1:18
office says it's reviewing the findings,
1:21
which is exactly what it said the first
