Hidden Risks in Georgia Senate’s AI Report

The Georgia Senate Study Committee on AI has released its recommendations, covering topics such as data privacy, deepfake technology, AI transparency, and law enforcement applications. While the report aims to be forward-thinking, certain proposals raise concerns about privacy, liberty, and potential government overreach.

The report reflects a familiar pattern in government regulations, where strict rules are often imposed on private entities, while comparable limitations on government are notably absent. This imbalance highlights critical concerns about fairness, accountability, and the safeguarding of individual freedoms, particularly when powerful emerging technologies are involved.

The report includes a variety of recommendations, but these are the ones that stood out to me as particularly worth discussing. Let’s break down these findings and explore their implications.

AI in Law Enforcement: A Gateway to Overreach

The report champions AI-enabled emergency response systems and encourages law enforcement agencies to adopt AI for predictive modeling and efficiency. While these tools may have legitimate applications, their misuse—and their dragnet nature—raise serious concerns.

The report’s recommendation to encourage the use of AI in public safety and law enforcement lacks the nuance and safeguards necessary to protect individual privacy and liberty. While the idea of AI-enabled emergency response systems and data-driven decision-making might sound beneficial, the lack of specific protections raises significant concerns about how this technology might be applied.

A “Blanket Greenlight”

The report essentially gives a greenlight for AI adoption in law enforcement without addressing potential abuses or overreach. There’s no mention of safeguards to protect citizens from invasive surveillance, misuse of data, or errors inherent in AI systems. This omission is particularly troubling given how AI has already been misused in policing and public safety contexts in other states and countries.

Predictive Policing: A Logical—and Concerning—Next Step

While the report doesn’t explicitly mention predictive policing, it does endorse “data-driven predictive models” for use by public safety.

The Committee also received testimony on “Predictive Response” by “Using historical data to identify potential hotspots to stage EMS + paramedics.” This approach is already practiced without AI and is undeniably beneficial. When seconds count, having EMS already in the vicinity can make a life-or-death difference compared to waiting for an ambulance to leave from a hospital or a distant depot.

However, there’s a key distinction: EMS respond when they are called—they don’t patrol neighborhoods looking for emergencies. Applying this same predictive behavior to policing introduces entirely different dynamics and risks.

While proactive EMS staging is a proven good, policing inherently involves authority, enforcement, and the potential for confrontation, which makes its application far more fraught with ethical concerns.

Predictive policing uses algorithms to forecast where crimes might occur or who might commit them, based on historical data. These systems have repeatedly been shown to reinforce existing biases in law enforcement. Neighborhoods flagged as “hotspots” garner more police attention, creating a feedback loop where increased police presence leads to more arrests and further skews the data.

The Privacy Problem

The glaring omission in the report is any mention of privacy protections. AI systems in public safety often rely on vast amounts of personal and public data—license plate readers, facial recognition, phone tracking, and more. Without strict guidelines on how this data is collected, stored, and used, such systems pose a serious risk to individual privacy.

For example, AI-powered emergency response tools might seem harmless on the surface, but they could inadvertently become part of a broader surveillance network. Imagine a system tracking mobile devices during a disaster that later gets repurposed to monitor citizen movements under other circumstances.

The Dragnet Problem

The report’s lack of safeguards is particularly troubling given the dragnet nature of many AI-powered law enforcement tools, which indiscriminately collect data on everyone, regardless of criminal suspicion.

Take Automated License Plate Readers (ALPRs) as an example. These systems don’t activate only during active searches for fugitives or emergencies. Instead, they are always running—scanning and recording the license plates of every passing vehicle. Over time, this creates a detailed log of people’s movements throughout a city, often without their knowledge or consent.

Such systems don’t discriminate between innocent individuals and those under investigation, turning public spaces into areas of constant surveillance. This approach treats everyone as a potential suspect, eroding the principle that individuals are presumed innocent until proven guilty. Over time, the aggregation of this data can lead to significant privacy violations, as well as opportunities for misuse by those with access to these systems.

Accountability and Oversight Are Key

If AI is to be used in law enforcement and public safety, the focus must be on accountability and oversight. Clear guidelines should limit the scope of AI applications, ensuring that the technology is used responsibly and transparently. Privacy protections must be front and center, with strict rules about data usage and robust mechanisms for citizens to challenge misuse.

The report’s endorsement of AI for law enforcement and public safety is concerning because it fails to address these critical issues. By giving a blanket greenlight without outlining protections, the committee risks enabling technology that could harm privacy and liberty more than it helps public safety.

Data Privacy Laws: A Trojan Horse?

By contrast, recommendations for private use of AI are far-reaching, placing significant regulatory burdens on businesses while leaving government use of similar technologies with far fewer constraints.

The report recommends Georgia adopt data privacy laws “similar to other states.” While privacy laws like California’s CCPA are marketed as safeguards for individuals, they often have unintended consequences. These laws can lead to centralized data control, giving governments the power to define and enforce privacy standards. This creates an environment ripe for abuse, where “protecting privacy” can be twisted into “monitoring compliance,” effectively increasing surveillance.

Privacy is best protected by empowering individuals to control their own data, not by giving bureaucrats the authority to dictate what privacy looks like. These laws, while perhaps well-intentioned, often prioritize government oversight over genuine freedom.

Banning Deepfakes: The Slippery Slope to Censorship

The committee’s recommendation to criminalize deepfakes for disinformation and coercion includes a sweeping statement: “Advertising, influencing, intimidating, or coercing individuals/entities through deep fake AI has no legitimate purpose and should be identified and banned with developers held accountable.” While no one would argue in favor of coercion or intimidation, laws already exist to address those acts, regardless of whether AI is involved. Why is an additional layer of criminalization necessary?

The Overreach into Advertising and Influencing

The real concern lies in the inclusion of “advertising” and “influencing” under the umbrella of illegitimacy. This broad phrasing raises many questions. Here are a couple that immediately came to my mind:

• Does this include AI-generated characters in advertisements? Using non-existent people generated by AI for ads or social media campaigns is already becoming a common practice. Such innovations can save costs, enhance creativity, and even protect the privacy of real individuals who might otherwise appear in these roles.
• What about deepfakes of oneself? Imagine a content creator or influencer using their own deepfake avatar to save time or reduce their on-camera presence. Would this be considered “illegitimate influencing” under such a law?

Without clear definitions, these rules risk penalizing benign or even beneficial uses of AI-generated content, stifling creative industries and forcing compliance burdens onto developers and advertisers who are otherwise operating ethically.

When Deepfakes Are Obviously Fake

Another issue is how laws like this fail to account for the audience’s discernment. Many deepfakes, especially in their current state, are clearly fake. The exaggerated, uncanny elements of the technology make it obvious to viewers that the content isn’t real. Satirical videos or humorous depictions using deepfake technology often fall into this category.

If a deepfake is so blatantly artificial that it poses no reasonable risk of deception, why should the government intervene? Most people are capable of assessing such content and deciding for themselves whether it’s legitimate or simply a creative use of AI. Imposing sweeping bans or holding developers accountable for such creations suggests a lack of trust in individual discernment.

Chilling Effects on Innovation and Expression

Labeling all advertising or influencing via deepfake AI as illegitimate could discourage technological advancements in AI and creative industries. Developers may shy away from building tools with legitimate applications for fear of being held accountable for misuse beyond their control. This chilling effect on innovation could hinder progress in areas like education, entertainment, and even accessibility technologies.

A Better Approach

While there are legitimate concerns about harmful uses of deepfakes—like impersonating someone to commit fraud or spread disinformation—any regulation must be carefully crafted to address specific harms without overstepping. Broad, vague language that criminalizes “advertising” or “influencing” risks conflating creative and legitimate uses with malicious acts.

Instead of banning entire categories of deepfake use, the focus should be on intent and impact. Laws should target cases where there is clear evidence of harm, deception, or coercion, while leaving room for creativity, innovation, and personal autonomy.

Deepfake technology is a powerful tool that, like any technology, can be used for good or ill. While addressing malicious uses is necessary, overbroad laws risk suppressing legitimate applications and stifling innovation. Rather than assuming the public can’t discern the fake from the real, we should trust individuals and regulate based on clear harm—not on potential misuse.

Transparency or Annoyance?

The committee’s recommendation to require full disclosure in any interaction between AI and humans might seem like a step toward transparency, but its practical implications could be counterproductive. If implemented poorly, this rule could make digital interactions as frustrating as the ubiquitous GDPR cookie notices that plague the web.

Initially designed to empower users, GDPR cookie pop-ups have done little more than annoy them, forcing people to click through yet another layer of bureaucracy before accessing content. Few users read these notices, and even fewer change their default settings, meaning their practical value is negligible. Instead of fostering informed consent, they have turned into a compliance box-checking exercise that diminishes user experience without enhancing privacy.

Similarly, requiring constant AI interaction disclosures could introduce unnecessary interruptions in day-to-day interactions. Imagine encountering pop-ups, warnings, or disclaimers every time you chat with a virtual assistant, use an automated chatbot, or engage with an AI-driven tool. Over time, these disclosures would likely be ignored, undermining their intended purpose while making digital services less user-friendly.

Transparency is important, but it must be implemented in a way that provides genuine value without burdening users or businesses. Overregulation risks turning helpful AI applications into a minefield of mandatory disclaimers, bogging down innovation and user experience alike. Instead of copying the worst aspects of GDPR, Georgia should seek approaches that prioritize clear, concise, and meaningful communication about AI without overwhelming users.

A State AI Board: Bureaucratic Theater or a Tool for Overreach?

The committee’s proposal to create a state board for artificial intelligence raises serious concerns about its purpose and potential impact. Historically, such government bodies tend to fall into one of two categories: toothless committees that waste resources or power-hungry entities that stifle progress and freedom.

Scenario A: The Do-Nothing Board

In its most benign form, the state AI board could amount to little more than a bureaucratic vanity project. This scenario would see the board filled with political appointees or disconnected officials who spend their time drafting reports, holding ceremonial meetings, and patting themselves on the back for “leading” in AI governance.

Such a board would likely have minimal impact on the AI ecosystem, serving only as a taxpayer-funded platform for resume-building and vague proclamations about ethical AI. While relatively harmless compared to the alternative, this setup would waste time and resources better spent elsewhere—resources that could have been used to support meaningful innovation or education about responsible AI use.

Scenario B: The Overreach Machine

The far more concerning scenario is a board granted real power. If such a body is empowered to regulate AI in Georgia, the potential for abuse and overreach becomes significant. With vague mandates like “ensuring transparency” or “protecting privacy,” the board could interpret its authority broadly, imposing burdensome regulations that stifle innovation.

Consider the ripple effects of an overzealous AI board:

• Regulatory Barriers to Innovation: Small businesses and startups, already operating on thin margins, could find themselves unable to navigate complex compliance requirements, leaving the AI space to large corporations with deep pockets.
• Political Agendas: Government boards often become tools of the prevailing political winds. An AI board could selectively enforce regulations based on ideological biases, chilling speech or innovation deemed “unacceptable” by those in power.
• Abuse of Power: Once established, such boards rarely limit themselves to their original scope. Over time, their authority could expand, enabling intrusive oversight into private sector activities and personal freedoms, all in the name of “protecting the public.”

Finding the Balance

The creation of any regulatory body should come with clear limitations and safeguards against abuse. Unfortunately, history shows that government entities seldom remain benign. Even a “do-nothing” board can morph into an overreach machine if given the chance.

Instead of creating a state AI board, Georgia could explore alternative approaches that decentralize power and foster voluntary standards for transparency and accountability. Private industry, educational institutions, and advocacy groups are often better equipped to address these issues without the baggage of bureaucracy or political agendas.

The risks of a state AI board—whether as a toothless waste of money or a tool for regulatory abuse—outweigh its potential benefits. Georgia should be cautious about introducing a body that could hinder progress and personal liberty in the name of governing AI.

Protecting Privacy in the Age of AI

Overall, the Senate Study Committee on AI’s recommendations present a mixed bag. While their intentions may be to harness AI for good, the proposals risk empowering the government at the expense of individual freedom. Privacy laws could lead to surveillance creep, deepfake regulations may stifle free expression, and AI in law enforcement could erode civil liberties.

Instead of creating new bureaucracies and regulations, the focus should be on decentralizing power, empowering individuals, and encouraging innovation without compromising liberty.

The fight for privacy and freedom in the age of AI and surveillance has never been more urgent. As tools like automated license plate readers, predictive policing models, and deepfake regulations grow in scope, it’s clear that technology is evolving faster than the safeguards meant to protect us. Without vigilance, these advancements risk eroding the very liberties they claim to serve.

At Banish Big Brother, we’re not just highlighting these issues—we’re helping individuals take meaningful action to protect their rights. That’s why we’ve created the 5-day “Take Control of Your Privacy” email course. This free course delivers practical steps you can take to secure your personal data, resist invasive technologies in your community, and push back against the growing surveillance state.

Each day, you’ll get clear, actionable advice—whether it’s securing your home network, protecting sensitive work data, or advocating for privacy laws that truly serve the public. Plus, you’ll receive the Banish Big Brother Toolkit, a free eBook packed with resources to help you stay one step ahead of surveillance under the guise of so-called “Smart Cities.”

Sign up today to get started. Privacy is worth protecting, and the first step starts here. Together, we can push back and reclaim our freedoms.