The Florida Senate
BILL ANALYSIS AND FISCAL IMPACT STATEMENT
(This document is based on the provisions contained in the legislation as of the latest date listed below.)
Prepared By: The Professional Staff of the Committee on Rules
BILL: CS/CS/SB 1680
INTRODUCER: Rules Committee; Judiciary Committee; and Senator Bradley
SUBJECT: Advanced Technology
DATE: February 14, 2024 REVISED:
ANALYST STAFF DIRECTOR REFERENCE ACTION
1. Collazo Cibula JU Fav/CS
2. Collazo Twogood RC Fav/CS
Please see Section IX. for Additional Information:
COMMITTEE SUBSTITUTE - Substantial Changes
I. Summary:
CS/CS/SB 1680 establishes the Government Technology Modernization Council, an advisory
council within the Department of Management Services, to generally advise the Legislature on
new technologies, artificial intelligence, and related issues. It also creates s. 827.072, F.S.,
entitled “Generated child pornography,” which makes it a crime to knowingly possess, control,
intentionally view, or create generated child pornography.
The general purpose of the advisory council is to study and monitor the development and
deployment of new technologies and provide reports on recommendations for the procurement
and regulation of such systems to the Governor and the Legislature. Accordingly, the bill
requires council members to meet at least quarterly and to perform several duties, including the
preparation and submittal of an annual report to the Governor and Legislature addressing the
modernization of government technology. The bill also provides for the composition of the
advisory council and regulates other aspects of service on the council.
The new criminal statute defines the terms “generated child pornography,” “intentionally view,”
and “sexual conduct,” and makes it a crime to knowingly possess, control, intentionally view, or
create generated child pornography. Each instance of possession, control, or intentional viewing
constitutes a separate offense. Anyone convicted of violating the statute is subject to up to five
years in prison and a $5,000 fine, as well as enhanced penalties under habitual offender statute.
The bill takes effect on July 1, 2024.
BILL: CS/CS/SB 1680 Page 2
II. Present Situation:
Artificial Intelligence
Generally
Artificial intelligence (AI) is the development of computer systems to perform tasks that
normally require human intelligence, such as learning and decision-making.1 It enables computer
systems to receive information that is either provided to them by others or gathered by them (e.g.
through camera lenses or other sensors), which they can then process and respond to in some
meaningful way. To a certain extent, AI systems are capable of adapting their behavior by
analyzing the effects of previous actions and working autonomously.2
Investments in AI have led to many of the transformative advancements that U.S. consumers rely
upon every day,3 including mapping technologies, voice-assisted smartphones, handwriting
recognition for mail delivery, financial trading, smart logistics, spam filtering, and language
translation. AI advances have also provided significant social benefits in areas such as precision
medicine, environmental sustainability, education, and public welfare.4
Types of AI
AI may be generally classified in one of three classes based on its capabilities or its
functionalities:5
Artificial Narrow AI. Also known as Weak AI, Artificial Narrow AI is the only type of AI
that exists today. All other forms of AI are theoretical. Machines using Weak AI can only
perform specific tasks using human-like capabilities. They can do nothing more than what
they are programmed to do. Examples of Artificial Narrow AI include Siri, Alexa, and
ChatGPT.6
General AI. Also known as Strong AI, General AI is only a theoretical concept. Any machine
or application using Strong AI in the future would be able to use what they have learned in
the past to accomplish new tasks in different contexts without the need for additional training
by human beings. In other words, they would be able to learn, perceive, understand, and
function completely like a human beings.7
Super AI. Also known as artificial superintelligence, Super AI is strictly theoretical. If ever
realized, machines using Super AI would think, reason, learn, make judgments, and possess
cognitive abilities surpassing those of human beings. Machines possessing Super AI
capabilities would have evolved beyond the point of understanding human sentiments and
1
National Conference of State Legislatures (NCSL), Artificial Intelligence 2023 Legislation, Jan. 12, 2024, https://www.ncsl.
org/technology-and-communication/artificial-intelligence-2023-legislation.
2
European Parliament, What is artificial intelligence and how is it used?, E.U. News, Jun. 20, 2023, https://www.europarl.
europa.eu/news/en/headlines/society/20200827STO85804/what-is-artificial-intelligence-and-how-is-it-used.
3
U.S. Department of State, Artificial Intelligence (AI), https://www.state.gov/artificial-intelligence/ (last visited Jan. 20,
2024).
4
Id.
5
IBM, Understanding the different types of artificial intelligence, Oct. 12, 2023, https://www.ibm.com/blog/understanding-
the-different-types-of-artificial-intelligence/; Naveen Joshi, 7 Types of Artificial Intelligence, Jun. 19, 2019, Forbes,
https://www.forbes.com/sites/cognitiveworld/2019/06/19/7-types-of-artificial-intelligence/?sh=7b5ddf4d233e.
6
Id.
7
Id.
BILL: CS/CS/SB 1680 Page 3
experiences to feeling emotions, having needs, and possessing beliefs and desires of their
own.8
Under the umbrella of Artificial Narrow AI or Weak AI, there are four kinds of AI based upon
functionalities:9
Reactive Machine AI. Reactive machines are AI systems with no memory. They are designed
to perform very specific tasks. They can only work with presently available data because
they cannot recollect previous outcomes or decisions. Reactive Machine AI stems from
statistical math and can analyze vast amounts of data to produce a seemingly intelligent
output. Examples of machines and applications that rely upon Reactive Machine AI include
IBM Deep Blue (IBM’s chess-playing supercomputer) and the Netflix recommendation
engine.10
Limited Memory AI. In addition to having the capabilities of purely reactive machines,
Limited Memory AI machines and applications are also capable of learning from historical
data to make decisions. Almost all present-day Limited Memory AI applications, including
Generative AI tools (e.g. chatbots and virtual assistants) and self-driving vehicles, are
Limited Memory AI machines and applications.11
Theory of Mind AI. Theory of Mind AI is a kind of General AI that exists in concept only at
this time. It is the “next level” of AI systems that researchers are currently developing.
Machines and applications using a Theory of Mind level AI will be able to understand the
thoughts and emotions of other entities. In theory, this will allow them to simulate human-
like relationships and to contextualize artwork and essays, which today’s Generative AI tools
are unable to do.12
Self-Aware AI. Self-Aware AI is a kind of Super AI that exists in concept only at this time. It
is strictly theoretical. If ever achieved, it will have the ability to understand its own internal
conditions and traits along with human emotions and thoughts. It will also have its own set of
emotions, needs, and beliefs.13
Generative AI
Generative AI is a type of Limited Memory AI technology14 that can produce high-quality
content, including text, images, audio, or video, within seconds when prompted by a user.15
Although it was first introduced in the 1960s, it was not until 2014, with the introduction of
8
Id.
9
Id.
10
Id.
11
Id.
12
Id. Emotion AI is a kind of Theory of Mind AI that is currently under development. Researchers hope that it will one day
have the ability to analyze voices, images, and other kinds of data to recognize, simulate, monitor, and respond appropriately
to humans on an emotional level. To date, Emotion AI is unable to understand and respond to human feelings. Id.
13
Id.
14
George Lawton, What is generative AI? Everything you need to know, TechTarget, Jan. 2024, https://www.techtarget.com/
searchenterpriseai/definition/generative-AI.
15
Government Accountability Office (GAO), Science, Technology Assessment, and Analytics, Science & Tech Spotlight:
Generative AI (June 2023), available at https://www.gao.gov/assets/gao-23-106782.pdf; George Lawton, What is generative
AI? Everything you need to know, TechTarget, Jan. 2024, https://www.techtarget.com/searchenterpriseai/definition/
generative-AI.
BILL: CS/CS/SB 1680 Page 4
generative adversarial networks, or GANs (a type of machine learning algorithm),16 that
Generative AI could convincingly create authentic images, videos, and audio of real people. 17
Generative AI systems learn patterns and
relationships from massive amounts of data,
which enables them to process and create
new content that may be similar, but not
identical, to the underlying training data.
Such systems rely upon sophisticated
machine learning algorithms and statistical
models to work.18
In order to generate new content, Generative
AI users are required to submit prompts that
guide the generation of new content. Many
iterations may be required to produce the
intended result because Generative AI is
sensitive to the wording of prompts.19 How Generative AI Works20
Because Generative AI can do so much, it has many potential applications, including in
education, government, medicine, and law. Applications include:
Writing a speech in a particular tone.
Summarizing complex research.
Assessing legal documents.
Creating images for different applications.
Composing music.
Composing poems.
Designing molecules for new drugs.
Generating programming codes.
Translating languages.
Implementing chatbots.
16
“A generative adversarial network (GAN) is a deep learning architecture. It trains two neural networks to compete against
each other to generate more authentic new data from a given training dataset. For instance, you can generate new images
from an existing image database or original music from a database of songs. A GAN is called adversarial because it trains
two different networks and pits them against each other. One network generates new data by taking an input data sample and
modifying it as much as possible. The other network tries to predict whether the generated data output belongs in the original
dataset. In other words, the predicting network determines whether the generated data is fake or real. The system generates
newer, improved versions of fake data values until the predicting network can no longer distinguish fake from original.”
Amazon Web Services (AWS), What is a GAN?, https://aws.amazon.com/what-is/gan/ (last visited Jan. 20, 2024). GAN can
generate images, training data for other models, complete missing information, and generate 3D models from 2D data. Id.
17
George Lawton, What is generative AI? Everything you need to know, TechTarget, Jan. 2024, https://www.techtarget.com/
searchenterpriseai/definition/generative-AI.
18
Government Accountability Office (GAO), Science, Technology Assessment, and Analytics, Science & Tech Spotlight:
Generative AI (June 2023), available at https://www.gao.gov/assets/gao-23-106782.pdf. Training data can include open-
source information, such as text and images from the internet. Id.
19
Id.
20
Id.
BILL: CS/CS/SB 1680 Page 5
Deploying “deepfakes.”21
Improving dubbing for movies.
Designing physical products and buildings.22
As of early 2023, emerging Generative AI systems have reached more than 100 million users and
have attracted global attention to their potential applications.23
The U.S. Government Accountability Office has identified several opportunities and challenges
in connection with the proliferation of Generative AI systems.24 With respect to opportunities,
Generative AI can quicken access to ideas and knowledge by helping people more efficiently
gather new information; help automate a wide variety of administrative and repetitive tasks; and
enhance the productivity of many industries.25 With respect to challenges, because Generative AI
systems can respond to harmful instructions, they can increase the speed and scale of many real
world harms, such as facilitating the development and proliferation of false information;
facilitating the use of copyrighted, proprietary, or sensitive data, without the owner’s or subject’s
knowledge; reducing privacy for users, including minors, through the retention of personally
identifiable information without consent; and facilitating the storage and use of sensitive
information by foreign adversaries.26
Regulation
Concerns about the potential misuse or unintended consequences of AI have prompted efforts to
examine and develop standards at the federal and state levels.27
For example, the White House Office of Science and Technology Policy has published a
document identifying principles that should guide the design, use, and deployment of automated
21
Deepfake AI is a type of AI used to create convincing images, audio, and video hoaxes. “Deepfakes” are created using a
combination of techniques, including face swapping. Algorithms learn the unique features of a person’s face, such as the
shape of his or her nose, the size of his or her eyes, and the position of his or her eyebrows, to create new images or videos
using the person’s likeness. The new image or video is very realistic and it may be difficult to tell that it has been
manipulated. Jerome Thiebaud, How Badly Will Deepfakes Weaponize Generative AI?, Centific, https://www.centific.com/
how-badly-will-deepfakes-weaponize-generative-ai/ (last visited Jan. 22, 2024).
22
Government Accountability Office (GAO), Science, Technology Assessment, and Analytics, Science & Tech Spotlight:
Generative AI (Jun. 2023), available at https://www.gao.gov/assets/gao-23-106782.pdf; George Lawton, What is generative
AI? Everything you need to know, TechTarget, Jan. 2024, https://www.techtarget. com/searchenterpriseai/definition/
generative-AI.
23
Government Accountability Office (GAO), Science, Technology Assessment, and Analytics, Science & Tech Spotlight:
Generative AI (Jun. 2023), available at https://www.gao.gov/assets/gao-23-106782.pdf.
24
Id.
25
Id.
26
Id.
27
NCSL, Artificial Intelligence 2023 Legislation, Jan. 12, 2024, https://www.ncsl.org/technology-and-communication/
artificial-intelligence-2023-legislation.
BILL: CS/CS/SB 1680 Page 6
systems.28 And the U.S. National Institute of Standards and Technology29 is holding workshops
and discussions with the public and private sectors to develop federal standards for the creation
of reliable and trustworthy AI systems.30
In 2023, at least 25 states, Puerto Rico, and the District of Columbia introduced AI bills, and 18
states and Puerto Rico adopted resolutions or enacted legislation. Examples include:31
Connecticut:
o Required its Department of Administrative Services to conduct an inventory of all
systems employing AI in use by state agencies and, beginning Feb. 1, 2024, to perform
ongoing assessments of such systems, to ensure that they will not result in unlawful
discrimination or disparate impact.
o Required its Office of Policy and Management to establish policies and procedures
concerning the development, procurement, implementation, utilization and ongoing
assessment of systems employing AI in use by state agencies.32
Louisiana adopted a resolution requesting its Joint Committee on Technology and
Cybersecurity to study the impact of AI in operations, procurement, and policy.33
Maryland established a grant program, its Industry 4.0 Technology Grant Program, to assist
certain small and medium-sized manufacturing enterprises with implementing new “industry
4.0” technology or related infrastructure. The definition of industry 4.0 includes AI.34