Friday, October 16, 2026, 10 a.m. (New York City time), Council Chambers - City Hall
Committee: Committee on Technology
The Committee on Technology, chaired by Council Member Carmen N. De La Rosa and joined by Speaker Julie Menin, held a follow-up oversight hearing on the risks posed by artificial intelligence, continuing the prior week's Committee of the Whole hearing; no bills were voted on. Gabriella Madrid of SpaceX's government affairs team testified under oath after the Council issued a subpoena, and members questioned her about Grok-generated sexualized images (including one the Speaker said she produced of herself that morning), antisemitic outputs, the size of the company's safety team, independent third-party evaluators, liability and insurance, whistleblower retaliation, medical uses and data deletion; she said she could not answer several questions and would get back to the Council. Former SpaceX employee Devin Kim and his attorney testified that he was fired after raising safety concerns, describing a safety team of about three people and urging whistleblower protections and broader incident reporting. Members of the public testified on privacy, third-party audits (Intro 1100), Brooklyn Navy Yard tenants, deepfakes and AI in emergency dispatch, and the Speaker closed by calling SpaceX's testimony not credible and unacceptable.
0:15 Morning and welcome to today's
0:15 New York City Council hearing
0:16 for the committee on
0:17 technology at this time I
0:19 would like to remind everyone
0:20 to silence all electronic
0:22 devices. Also at this point
0:23 going forward and no one is
0:25 supposed to approach the days
0:32 she begin. Good morning,
0:33 everyone. I'm New York City
0:35 council speaker Julie Menin
0:36 .want to welcome everyone t
0:37 .the technology chair council
0:44 member Carmen. We look forward
0:45 to having a very fulsome
0:47 hearing and this is a sequel
0:50 to our initial hearing on a
0:51 aye today the New York City
0:53 council continues a crucial
0:56 conversation that continues
0:57 from last week's committee of
0:59 the whole with the world's
1:01 largest. Aye, aye. Cos we were
1:04 the first legislative body I
1:05 nthe courynt that heard from
1:07 the leading companies on their
1:08 models that have captured as
1:10 much attention as they have
1:11 concern across the globe as we
1:14 heard from both experts and
1:17 the public, there is good
1:18 reason to remain apprehensive
1:20 about some of the safety risks
1:23 around artificial intelligence
1:23 as well as the companies
1:25 behind these models. The
1:27 answer is quite frankly, that
1:28 we received from last week's
1:31 hearing only reinforced the
1:33 necessity for responsible
1:36 safeguards to prevent harm and
1:38 the idea that the companies, I
1:41 would not would self regulate
1:42 is not a concept that we
1:44 support at all at the same
1:47 time we want to ensure that we
1:49 are not stifling innovation
1:50 .New York City is rapidly
1:52 growing as not only a tec
1:55 hcapital, but also the capital
1:57 of the world with over 40000
1:59 0jobs more than the City of san
2:00 francisco, I might add and we
2:03 understand and we want to
2:05 support keeping those jobs
2:06 here in New York supporting
2:08 cutting edge research
2:10 supporting startups located
2:11 right here in our great City,
2:14 so we know that no matter how
2:17 large or how small these
2:20 various companies are we know
2:21 that those companies are vital
2:23 to our City in terms of
2:25 cutting edge research medical
2:28 breakthrough scientific
2:29 discoveries and boosting
2:31 economic growth, but we canno
2:34 t
2:34 . Wait for these companies t
2:36 oact responsibly when we see
2:37 that some of these models are
2:39 able to hack into other
2:41 companies and also into
2:44 government websites as well as
2:45 we work to ensure that there
2:48 are responsible, suitable
2:50 safeguards in place following
2:51 alarming public reports on, we
2:55 want to welcome spacex to thi
2:56 shearing. It is not only New
2:58 yorkers who want to hear your
3:00 testimony today, but people
3:02 across the world. So we thank
3:04 them for being here under oath
3:07 to answer our questions about
3:09 this evolving technology like
3:10 I said at the hearing last
3:13 week. These companies are
3:14 developing technology that can
3:15 reshape so many aspects of our
3:19 lives and yet they are being
3:20 led, oftentimes by individuals
3:22 who are publicly warning about
3:24 the risk with their models. So
3:26 we deserve to have those risks
3:29 , so safety concerns answered
3:32 as well as a discussion about
3:33 what are the best legislative
3:36 solutions to address those
3:38 risks this City Council has
3:39 proposed solutions to address,
3:43 for example, emergency
3:44 preparedness. We have bills to
3:46 require a party party
3:49 independent validation that is
3:50 free from conflict of interest
3:52 we have a bill first in the
3:54 nation to incentivize
3:55 whistleblowers to come forward
3:58 as well as a bill to devise a
3:59 plan for the scenario in which
4:01 , in case a. I hacks into City
4:04 systems as well as other bills
4:06 as well. I want to end by
4:08 saying that at our hearing
4:10 last week we showed that New
4:12 York City has demonstrated t
4:14 othe world that cities can and
4:16 should take the lead on a I
4:19 safety our federal governmen
4:20 tis simply failing to act. Of
4:23 course we would like to see
4:25 regulation at the federal
4:26 level and it is our fervent
4:28 hope that by putting forward
4:30 these legislative ideas that
4:32 it will ultimately lead to the
4:34 federal government coming up
4:36 with national safety standards
4:38 and regulations of a I just as
4:41 I said last week. We do not
4:42 leave it to the pharmaceutical
4:44 industry to self regulate. We
4:47 have the we do not leave it to
4:48 the airline industry to self
4:50 regulate we have and an
4:52 independent safety board when
4:55 there are safety concerns we
4:58 have the sec that regulates
5:00 the financial industry with a
5:02 I what we are looking for is
5:06 responsible safeguards to put
5:07 in place New York City has and
5:10 always will be a leader in
5:11 addressing the key issues that
5:13 affect New yorkers and a I is
5:14 simply no different. I look
5:17 forward to today's
5:18 conversation. I want to thank
5:19 our team, our staff for
5:21 helping to organize this
5:23 important hearing and now I
5:23 want to turn it over to chair
5:25 delarosa for a statement.
5:26 Thank.
5:27 You. Thank you, speaker
5:29 .mennen. Good morning, everyone
5:30 . I'm Council Member carme
5:31 nrosa and I am the chair of the
5:33 committee on technology.
5:34 Welcome to our hearing on
5:36 continuing discussions about
5:37 the risk posed by artificial
5:40 intelligence.
5:41 Last week in this very same
5:42 room in a meeting of the
5:43 committee of the whole we
5:45 heard testimony from nearly
5:47 100 witnesses, including
5:48 representatives from opening I
5:50 metre and tropic google their
5:53 former employees tech
5:55 technical experts, academics,
5:57 elected officials and members
5:58 of the public.
5:59 Despite their different
6:00 points of view their message
6:02 was more or less consisten
6:06 tthe pace and progress of
6:08 frontier models does pose a
6:10 risk
6:10 . Some called for frontier
6:12 developers to slow.
6:13 Down, others focused on
6:15 alignment and control
6:18 mechanisms
6:18 . Warnings of super
6:19 intelligence mask surveillanc
6:21 e, energetic policing as well
6:23 as political bias and
6:25 discriminatory tendencies,
6:27 several witnesses raised
6:28 concerns about the loss of
6:30 motor control and others
6:32 warned about the potential
6:32 loss of jobs. However, much of
6:35 the testimony return to a
6:37 common set of recommendations
6:39 . Transparency from a
6:42 developers credit credible
6:43 safety measures. Third party
6:45 evaluators and protection for
6:49 personal data today, we plan
6:50 to continue this conversation
6:51 with another major company,
6:53 space x s I. We want to reap,
6:56 reap. We want to reap the
6:57 benefits of the eye so long as
6:59 these benefits come wit
7:00 hresponsible development and do
7:01 not cost us our safety, our
7:04 dignity and our civil rights.
7:07 Be a powerful tool for
7:10 efficiency and innovation in
7:12 science, medicine and
7:15 productivity at the same time
7:16 it introduces important risks,
7:18 including bias discrimination
7:20 ,job displacement sensors,
7:24 digital forgeries, fakes and
7:25 concerns that become even more
7:29 serious as agents become more
7:33 capable
7:33 . We should make sure tools
7:34 work as intended that we
7:35 understand how they work, how
7:37 they were trained where they
7:39 eopportunity to contest this
7:44 made by machines, a framework
7:47 where the very same people who
7:48 build a tools and have
7:51 competing motives are also
7:52 expected to regulate
7:55 themselves is wholly
7:56 insufficient intervention is
7:57 necessary from federal, state
7:58 and municipal governments t
8:01 oprotect our rights.
8:02 Once again, I look forward
8:04 to hearing more from industry
8:06 experts about what solutions
8:07 are possible to ensure New
8:09 yorkers can fully benefit fro
8:11 ma technological advancement
8:14 without compromising civil
8:14 rights, safety and privacy
8:17 . I now like to thank the
8:18 technology committee staff,
8:20 legislative council. Irene buy
8:21 off ski and policy analyst
8:22 Eric brown as well as my chief
8:23 of staff James burke,
8:26 legislative director. He by
8:28 mod comes director free
8:29 familiar and speaker man his
8:30 entire team for the continued
8:32 work on this important matter,
8:38 I'd also le tiko recognize
8:39 some of the council members
8:40 that have joined us,
8:44 councilmember Maloney, and
8:45 that's it for now more to come
8:47 . I now turn it back to
8:49 speaker mennen and the
8:50 committee council. Thank
8:51 . You so much chair. Delarosa,
8:52 I'm now going to pass it over
8:56 to counsel to do the
9:00 affirmation. Christina madrid.
9:01 You answer yes. After the
9:02 question, do you affirm to
9:03 tell the truth, the whole
9:04 truth and nothing but the
9:06 truth in your testimony before
9:09 this.
9:10 Committee? Yes. Thank you.
9:10 You begin your testimony
9:15 , Madam speaker charla Rosa.
9:15 And members of the committee
9:17 on technology. Thank you for
9:19 the opportunity to participate
9:20 today. My
9:21 . Name is gabriella madrid and
9:22 I am part of the government
9:23 affairs team at spacex. My
9:26 team's responsibility is to
9:28 maintain a close relationship
9:29 between the teams building our
9:31 mind advanced technologies,
9:33 including from models and the
9:34 communities that use and
9:35 benefit from them. Spake sex
9:38 has grown from a startup up
9:39 nearly 25 years ago to a
9:41 company launches more rockets
9:43 and orbits more satellites
9:44 every year than any other
9:45 nation or company developing
9:49 rapidly reusable rockets
9:50 . Has enabled spacex to
9:52 dramatically lower the cost of
9:53 lunch and make New nasa
9:55 missions and science possible,
9:58 including regular and reliable
9:58 crew and cargo missions to the
10:00 international space station
10:01 and the upcoming are chemist
10:03 mission to the moon are
10:06 starling satellites now make
10:08 possible reliable broadband
10:09 connectivity across over 160
10:11 countries, including to
10:12 locations where high speed
10:14 internet has never been
10:17 available before spacex
10:17 mission is to make humanity
10:21 multi planetary
10:21 . Earlier this year, spacex
10:23 acquired, blending spacex deep
10:26 technical expertise building
10:27 large, complex infrastructure
10:29 and complex space systems with
10:30 the development of super
10:31 intelligence and frontier
10:35 models.
10:35 Just as we have done with
10:36 space launch and human space.
10:39 Spacex put safety first
10:42 . As a mother. I know the
10:43 risks.
10:43 New yorkers worry about are
10:44 often those close to home,
10:47 protecting our children
10:49 preventing harmful content and
10:50 stopping this use of everyday
10:53 tools, and I can assure you
10:54 that safety is core to how we
10:57 build rock from pre training
10:58 through deployment. Our
11:00 approach starts with a simple
11:01 idea
11:02 . A super intelligence system
11:03 built to see the truth and
11:05 understand the universe acid
11:07 actually is will be more safe
11:09 ,reliable and useful to the
11:10 people who rely on it. Our
11:13 safety work is how we make
11:14 sure that hopes up in practice
11:16 by focusing on control
11:19 prevention, mitigation,
11:19 analysis and correction
11:22 concretely that effort begins
11:24 well before a model is
11:25 released for public use. We
11:27 train our models to refuse
11:29 assistance with dangerous
11:30 activity
11:31 . Evaluated for risks
11:32 including misuse of our terms
11:33 of service
11:34 . Attempts to get aroun
11:34 dsafeguards and lots of control
11:38 . We monitor systems after the
11:39 appointment, and we do not
11:40 release models we relieve are
11:42 insufficiently robots,
11:43 aren't sufficiently robust
11:44 against by bad actors or
11:46 follow minimal human control.
11:50 These practices are I lined in
11:51 our publicly available
11:52 frontier artificial
11:53 intelligence framework, along
11:54 with each New model card,
11:55 which describes how we assess
11:58 and reduce these risks across
11:58 the model's life cycle. We
12:02 also perform a full risk
12:04 assessment of our practices
12:05 and controls at least once per
12:08 year
12:09 . We relieve transparency is
12:10 fundamental to creating this a
12:12 safer ecosystem when something
12:13 goes wrong. We investigated
12:15 work to fix it and reported to
12:17 the authorities have went
12:20 tired. Use well as I can help
12:20 improve the lives ofhe t
12:22 people of New York by speeding
12:24 up research making public
12:26 information in syria to fin
12:27 dand use and helping agencies
12:29 and residents with complex
12:33 tasks, those benefits only
12:33 matter if the systems remain
12:35 safe and accountable
12:38 . This is why we continue to
12:39 invest in evaluation,
12:40 monitoring and human control
12:42 and why
12:42 . We want to work with policy
12:45 unpractical.
12:45 Evidence based protections
12:48 .I look forward to answering
12:49 your questions today and
12:50 continuing the conversation.
12:54 Thank you
12:54 , thank you very much for you
12:55 r
12:55 . Testimony, ms, we have a
12:57 number of questions
13:00 . On September 12, an article
13:03 entitled we must face a french
13:05 here stating that within 6 2
13:08 12 months a swarm of a I
13:11 agents quote could be capable
13:12 of taking over the entire
13:13 internet with a persistent
13:16 botnet, potentially causing
13:17 hundreds of billions of
13:17 dollars in damage end quote
13:19 . Was he correct here.
13:24 Um, I wouldn't. I'm not
13:25 familiar with that statement
13:26 in particular what I will say
13:28 is safety ist cord to spacex
13:30 and that goes from treat
13:32 retraining through released so
13:34 people stay in control of our
13:35 systems at every stage we do
13:37 not release models. We don't
13:38 not believe are safe and that
13:41 we cannot keep under
13:42 meaningful human control and
13:43 if.
13:43 A model does not mean that
13:44 bar, we hold it back and
13:45 strengthen the model and safe.
13:48 We train models to ensur
13:49 ethey don't assist with
13:50 dangerous activity, and I
13:51 evaluate them for misuse. S
13:55 othen why did elon musk state
13:57 on x and a post and responds
14:01 to those comments is right
14:05 . Um, I really we're
14:06 optimistic about. We obviously
14:08 take the risk seriously. Ellen
14:09 has been one of the earliest
14:11 and most prominent voices and
14:12 warning about the risk and
14:13 calling for oversight
14:16 . Um, but we are extremely
14:17 positive about what can be
14:19 achieved, we believe that
14:20 super intelligence can hel
14:21 ppeople solve problems that
14:23 were previously beyond reach,