As AI-powered hiring tools become more widespread, it is important to remember that some states have acted early and their laws should be reflected in an organization’s governance, risk, and compliance program. Maryland’s Labor and Employment Code Section 3-717 which became effective October 1, 2020, is one example. We break down the key requirements and highlight multistate considerations.

Q: Who does the law apply to?

The law applies to employers who use facial recognition services during applicant interviews. Its protections run to applicants, generally individuals who are interviewing for employment. The law does not, by its terms, extend to current employees or to other stages of the employment relationship beyond the interview. It is unclear, however, whether the law may also protect employees of an organization applying for other positions at the same organization.

Q: What does Maryland Section 3-717 actually prohibit?

The statute prohibits an employer from using a “facial recognition service” to create a “facial template” during a job applicant’s interview, unless the applicant has first provided consent. A “facial recognition service” is defined as technology that analyzes facial features and is used for the recognition or persistent tracking of individuals in still or video images, and a “facial template” is the machine-interpretable pattern of facial features extracted from one or more images of an individual by such a service.

Q: How does an applicant give consent under the Maryland law?

Consent must be provided through a signed waiver. The waiver must be written in plain language and include: (1) the applicant’s name, (2) the date of the interview, (3) a statement that the applicant consents to the use of facial recognition during the interview, and (4) whether the applicant has read the consent waiver.

Q: Does Section 3-717 apply to all uses of AI in hiring?

No. The Maryland statute is narrowly focused on facial recognition services used to create facial templates during interviews. It does not broadly regulate other AI-based tools an employer might use in the hiring process, such as resume-screening algorithms or chatbot-based assessments, unless those tools incorporate facial recognition technology as defined in the statute.

Q: How does the Maryland law differ from the Illinois Artificial Intelligence Video Interview Act (AIVIA)?

Although both laws regulate the use of technology during applicant interviews, there are several notable differences:

  • Scope of technology covered. Maryland’s Section 3-717 targets only “facial recognition services” used to create a facial template. The Illinois AIVIA, which became effective January 1, 2020, is broader: it applies whenever an employer asks applicants to record video interviews and uses “artificial intelligence analysis” of those videos, which can encompass a wider range of AI evaluations beyond facial recognition alone.
  • Disclosure and transparency. The Illinois AIVIA requires employers to (a) notify the applicant before the interview that AI may be used to analyze the video and assess their fitness for the position, and (b) provide an explanation of how the AI works and what general types of characteristics it evaluates. Maryland’s statute has no comparable pre-interview disclosure or transparency requirement — it requires only a signed consent waiver.
  • Form of consent. Maryland requires a specific written waiver containing the applicant’s name, interview date, a consent statement, and an acknowledgment of whether the applicant read the waiver. Illinois requires consent to be obtained before the interview but does not prescribe the same formalized waiver requirements.
  • Video sharing and deletion. The Illinois AIVIA includes additional protections that have no counterpart in the Maryland statute. Under the Illinois law, employers may not share applicant videos except with persons whose expertise or technology is necessary to evaluate the applicant, and employers must delete interview videos (including backup copies) within 30 days of an applicant’s request. Maryland’s Section 3-717 does not address video sharing or data deletion.

Q: What should multi-state employers take away from these laws?

An ongoing challenge for organizations operating in multiple states is the patchwork of similar laws addressing similar technologies, but sometimes with a different scope or context, along with some unique provisions.

Here, we focused on two laws that address the interview process for job applicants. One might ask whether there are other statutes that regulate the interview process in this way. Focusing only on regulation of job interviews might cause one to miss other critical compliance requirements.

The Maryland and Illinois laws discussed above may involve the collection of biometric information that, for example, also is protected under more general laws, such as the California Consumer Privacy Act (CCPA), the Illinois Biometric Information Privacy Act (BIPA), and other states with protections for such information (e.g., Colorado and Texas).

Additionally, neither the Maryland nor Illinois job interview laws includes obligations to safeguard the facial scan data collected during covered interviews. However, other state laws may include such a requirement. Maryland’s own Personal Information Protection Act (PIPA) requires:

a business that owns, maintains, or licenses personal information of an individual residing in the State [to] implement and maintain reasonable security procedures and practices that are appropriate to the nature of the personal information owned, maintained, or licensed and the nature and size of the business and its operations.

Under the PIPA, personal information includes biometric information.

It is not enough to think about just the activity the organization is engaged in – a job interview – employers need to consider a range of other issues to fully appreciate the regulatory environment for that activity – the kind of data collected, the location of the collection, who is collecting it, the state of residency of the person providing the information, how that data is collected and analyzed, among other things.

On September 2, 2026, Delaware’s Governor signed House Bill (HB) 380 and HB 381. HB 380 amends the Delaware Personal Data Privacy Act (DPDPA), which was enacted in 2023 and became effective January 1, 2025. HB 381 separately amends Delaware’s computer security breach notification law.

In short, what changes were made to the Delaware privacy and data-breach laws?

Together, the bills expand the businesses and data covered by existing law, increase protections for sensitive data, impose new vendor-management and automated decision-making requirements, and strengthen breach-notification obligations.

When do the changes take effect?

The HB 380 amendments to the DPDPA take effect January 1, 2027, while the HB 381 amendments to the breach-notification law take effect upon signing.

Which organizations does the amended DPDPA cover?

HB 380 lowers the DPDPA’s applicability thresholds. The law previously covered businesses that control or process the personal data of at least 35,000 consumers during the preceding calendar year. That threshold is now reduced to 10,000 consumers. For businesses that derive more than 20% of their gross revenue from selling personal data, the alternative threshold drops from 10,000 to 5,000 consumers.

The amendments also cause the DPDPA to apply to third parties that acquire personal data from a controller. Recall that third parties are defined broadly to include, with respect to personal data controlled by a controller, any person other than the relevant consumer, the controller of such personal data, or a processor or an affiliate of the processor or the controller.

Accordingly, organizations that fell below the original thresholds and potential third parties should reassess whether they are now subject to the DPDPA under the lower thresholds and changes enacted by HB 380. However, the changes also clarified organizations and categories of data that are excluded from the DPDPA, largely affecting financial institutions and the healthcare sector.

How did consumer rights change?

Consumer rights under the DPDPA have expanded in some significant ways. Consumers can now confirm whether controllers are including inferences about the consumer derived from personal data and processing personal data for profiling purposes to make significant decisions about the consumer.

Additionally, under the prior law, consumers had the right to obtain a list of “categories of third parties to which the controller has disclosed” their personal information. Under HB 380, consumers can obtain the list of the actual third parties, subject to several exceptions, including if the list would reveal a trade secret or that list cannot be compiled with reasonable effort. However, in that case, the controller must disclose all third parties to which it disclosed personal information.

The amendment also limits the consumer right of access with respect to certain sensitive data elements. For example, controllers may not disclose a consumer’s SSN or account password, but may only inform the consumer that the controller possesses that information.

Did the contents of the controller privacy notice change?

Yes, privacy notices must identify the controller for contact purposes, and include a description of the consumer’s personal data rights.

What new categories of sensitive data does HB 380 add?

HB 380 expands the definition of sensitive data to include national origin; health treatment or status; neural data; certain financial-account information and credentials; and government-issued identification numbers. It also covers inferences drawn from other information when those inferences are used to reveal or identify a sensitive characteristic.

What are the new requirements for processing sensitive data under Delaware law?

Processing sensitive data requires the consumer’s consent and must be reasonably necessary and proportionate to the disclosed purpose.

Are there restrictions on selling sensitive data?

Yes. A controller may sell sensitive data only when the disclosure is strictly necessary to provide or maintain a product or service affirmatively requested by the consumer. The controller must provide a clear and conspicuous notice, obtain the consumer’s consent, and retain a record of that consent for five years.

What new vendor-management obligations does HB 380 impose?

Controllers must enter into contracts with third parties receiving personal data. Those contracts must identify limited and specific purposes for the data, require the third party to provide the same level of privacy protection as the controller, and permit the controller to address unauthorized uses. Controllers must also conduct reasonable due diligence regarding third parties, including, at minimum, questionnaires and reviews of relevant documents.

What does HB 380 require for automated profiling and decision-making?

HB 380 adds requirements for profiling and reports used in decisions that produce legal or similarly significant effects, such as decisions about employment, housing, lending, insurance, education, or healthcare. When a report contributes to an adverse action, contractual provisions must require notice to the affected Delaware resident, identification of the data relied upon, and information about requesting human review where technically feasible.

Are impact assessments required for automated profiling?

Yes. Controllers engaging in covered automated profiling must conduct and document impact assessments. These assessments must address risks, inputs, outputs, performance measures, limitations, transparency, and post-deployment safeguards.

Does the employee-data exclusion still apply?

The existing employee-data exclusion is narrowed for personal data used in the context of profiling or disclosing to a third party a report for use in connection with any decision that produces legal or similarly significant effects on a resident of Delaware. In such cases, controllers and third parties have several requirements, including notice requirements to the residents concerning adverse actions, contract obligations, and additional consumer rights requests relating to personal information use in profiling.

What changes does HB 381 make to Delaware’s breach-notification law?

Like many states, the Delaware breach notification law provides an alternative way to provide notice (called “substitute notice”) when the cost of notification will exceed $75,000, the number of affected Delaware residents exceeds 100,000, or the organization does not have sufficient contact information to provide notice. Substitute notice typically involves email notice, conspicuous website notice, and notice to major statewide media. HB 381 updated the substitute notice requirement to also include notice to the Delaware Attorney General.

Additionally, in situations when an organization, through reasonable diligence, cannot identify within 60 days that personal information of Delaware residents was involved in a breach, it must provide required notification as soon as practicable after it determines such information was involved. HB 381 adds that notice to the Attorney General also must be provided within 60 days of that determination.

Does compliance with HIPAA or the Gramm-Leach-Bliley Act satisfy Delaware’s breach-notification requirements?

HB 381 narrows the existing exemption for entities regulated under laws such as HIPAA or the Gramm-Leach-Bliley Act. In short, compliance with regulator-established breach procedures now satisfies only Delaware’s 60-day timing requirement, rather than the entire state breach-notification law.

Who enforces the amended DPDPA?

The DPDPA amendments retain enforcement by the Delaware Department of Justice. The amendments do not create a private right of action.

What steps should businesses take to prepare for the new requirements?

Businesses should reassess whether they are covered under the lower applicability thresholds, inventory sensitive data and automated decision-making activities, update incident-response procedures, and review privacy notices, consent processes, third-party contracts, and vendor due diligence practices.

On June 4, 2026, Connecticut Governor Ned Lamont signed Public Act No. 26-73 (SB 472), “An Act Concerning the Electronic Surveillance of Employees,” which repeals and replaces Conn. Gen. Stat. § 31-48d. The law takes effect October 1, 2026, and significantly updates Connecticut’s longstanding requirements governing electronic monitoring of employees by employers.

Connecticut has regulated workplace electronic monitoring since 1998, and presently, several states have some form of notice, acknowledgment, and/or consent requirement for organizations engaging in electronic monitoring of their workforce.  These states include California, Delaware, Maine, New Jersey, and New York.

Before discussing the updated notice requirements, it is important to note that the new law does not change the definition of electronic monitoring. It remains:

“Electronic monitoring” means the collection of information on an employer’s premises concerning employees’ activities or communications by any means other than direct observation, including the use of a computer, telephone, wire, radio, camera, electromagnetic, photoelectronic or photo-optical systems, but not including the collection of information (A) for security purposes in common areas of the employer’s premises which are held out for use by the public, or (B) which is prohibited under state or federal law.

The use of cameras or the tracking of activities and communications on a company’s information systems quickly come to mind. However, the rapid expansion of AI technologies and the various use cases also should be assessed. Consider AI tools and use cases such as AI transcription and note taking tools, chatbots, and performance management platforms. These and other tools and use cases may require employers to revisit their notice and posting obligations under this expanded law in Connecticut.

What also remains the same is the definition of employee:

“Employee” means any person who performs services for an employer in a business of the employer, if the employer has the right to control and direct the person as to (A) the result to be accomplished by the services, and (B) the details and means by which such result is accomplished.

Updated Prior Written Notice and Posting Requirements

The updated monitoring law enhances the existing prior written notice requirements for employers engaging in electronic monitoring. Beginning October 1, 2026:

  • in addition to providing prior written notice to all employees who may be affected by electronic monitoring and specifying the types of monitoring, employers must inform those employees of the specific locations on the employer’s premises where such monitoring may occur; and
  • the existing obligation to post a notice of electronic monitoring in a conspicuous place which is readily available for viewing by employees, must also be posted in the specific location on the employer’s premises where the monitoring may occur. That posting also must describe the specific locations on the employer’s premises where the monitoring may occur.

Additionally, for employees hired on and after October 1, 2026, employers engaged in electronic monitoring must provide to such employees (prior to commencement of employment) a plain language written statement advising which activities are prohibited and may be monitored without giving prior written notice. Such activities include employees engaged in conduct which (i) violates the law, (ii) violates the legal rights of the employer or the employer’s employees, or (iii) creates a hostile workplace environment, and for which electronic monitoring can produce evidence of this misconduct.

Notably, the requirement to disclose the specific location of the premises being monitored does not apply where the premises is an airport, or when the employer has reasonable grounds to conduct such monitoring for security and employee safety purposes.

The updated monitoring law did not change the enforcement provisions. The Labor Commissioner may impose a civil penalty on any person that violates the notice and posting requirements. The maximum penalties are $500 for the first offense, $1,000 for the second, and $3,000 for the third and subsequent offenses.

Connecticut employers should review their monitoring activities, including deployments of AI tools and their respective use cases, and assess compliance with applicable posting, notice, and/or acknowledgment requirements in Connecticut and other states as applicable.

A family’s ordinary afternoon was shattered when an unknown number added multiple relatives, including a teenager, to a group chat and shared what appeared to be AI-generated explicit images of a family member, the Wall Street Journal (WSJ) reported. The goal was simple: to extort money by threatening to release the images more widely. The scheme was sophisticated, leveraging publicly available social media photos and AI tools to fabricate convincing pornographic content and to map out family relationships for maximum emotional impact.

Why This Matters for Your Organization

Sextortion, generally speaking, is the use of real or fabricated explicit images to coerce victims into paying money. According to reports, the practice is accelerating, and AI is the catalyst. Scammers no longer need to manually research their targets. Open-weight AI models running on dedicated machines can scrape social media profiles, analyze public data, and piece together personal relationships at scale.

What does this mean for organizations?  Whether the target is the employee or the organization, the organization could face significant disruption to the business, distracted employees, and reputational harm. Reported tactics by online extortionists include threats to post or send explicit imagery of employees in places online designed to be seen by the organization and its customers, such as social media platforms or, as in the scenario reported in the WSJ, by group text. Much the same as threat actors who plan spear phishing attacks, sextortionists do their research in order to understand and leverage their victims’ connections. The result could be a demand to the organization for large sums paid in cryptocurrency, in exchange for keeping the images fake though they may be, out of the public eye.

In addition to facing an extortion demand and the risk of harm to the brand and reputation, organizations also need to consider the impact on the victim employee(s). These kinds of attacks can cause the subject employees severe emotional distress distracting them from their job. The employee’s personal and business accounts, including email, also may have been compromised, helping the threat actor to expand the impact of the attack.

Are You Prepared?

Many organizations have robust cybersecurity protocols for protecting corporate systems, but far fewer address the human side of digital exploitation. Here are the questions every organization should be asking today:

  • Do your employees know what sextortion looks like? Many people still assume they aren’t targets because they aren’t wealthy or famous. The reality is that these AI-driven scams are automated and opportunistic— no one is too small a target. Consider updating your training programs to address this threat.
  • Does your security awareness training cover AI-generated threats? Traditional phishing training may not address the emotional manipulation involved in sextortion or the use of deep-fake technology to create convincing explicit images.
  • Do your employees know what to do if it happens to them? Experts, including the FBI, advise victims not to pay, to block and report the scammer to the FBI’s Internet Crime Complaint Center, and to immediately change passwords and end unknown device sessions. Does your organization have clear guidance or an employee assistance program that covers this type of incident?
  • Are you encouraging good digital hygiene? Simple steps—setting social media accounts to private, using strong and unique passwords, and enabling two-factor authentication—can significantly reduce exposure.
  • Does your incident response plan address this threat? Preparedness is critical for any significant business threat. In addition to ensuring there is a clear path for employees to report these incidents confidentially, whether they occur through personal or corporate channels, businesses should expand their incident response plan. They also should practice how they might respond to such an attack, which would know knowing what resources they have to provide support for the organization and affected employees. 

The Bottom Line

AI is supercharging old scams into something faster, more convincing, and more invasive. Sextortion is no longer a risk that only affects teenagers or public figures. It can reach anyone through any group chat, at any time. Organizations should consider this risk and improve preparedness.

A broad coalition of artificial intelligence developers, cybersecurity companies, financial institutions, technology providers, and other organizations has issued an open letter calling for a coordinated effort to strengthen cyber defenses.

“We have a limited window to strengthen cyber defenses.”

The letter warns that AI-enabled cyberattacks have become more widespread and sophisticated. At the same time, it argues that advances in AI can help defenders identify vulnerabilities, remediate weaknesses, and respond to incidents more efficiently. Importantly, the letter points to several industries facing significantly higher levels of risk for organizations in those industries and the sometimes many thousands or millions of people they serve, e.g., critical infrastructure, manufacturing, supply chain, utilities (water), health care.

The signatories propose three overarching principles:

  1. Recognize that existing security practices may no longer be sufficient;
  2. Expand access to AI-enabled defensive capabilities; and
  3. Build a collective response involving businesses, technology providers, governments, and frontier AI companies.

While not a new recommendation, the letter calls for cybersecurity to become an “immediate leadership priority.” It advocates that particular attention be given to high-risk vulnerabilities, least-privilege access, strong authentication, layered defenses, and security standards for technology that organizations purchase, develop, or deploy. In short, the letter suggests that the measures for achieving the standard of “reasonable safeguards” may have changed as the threat landscape rapidly evolves, and so too must the measures for safeguarding assets from those threats.  What constitutes “reasonable safeguards” is now a fast-moving target.

For those responsible for cybersecurity preparedness and incident response, the letter reinforces an important point: Organizations should not view AI defense as simply a technology-acquisition project. It is a governance, legal-risk, and operational-resilience issue. New tools will not compensate for unclear responsibility, incomplete asset inventories, weak vendor oversight, or incident-response plans that have never been tested. Organizations also should evaluate AI security tools carefully, including what data the tools collect, where that data is stored and how it is safeguarded, how model outputs are validated, and whether contracts appropriately allocate responsibility for security incidents.  

Taking these steps, among others, will help organizations make the best use of this limited window to strengthen their cyber defenses, positioning them as less vulnerable targets for AI-powered cyberattacks and to defend themselves against the class action lawsuits that are likely to become even more prevalent as these attacks gain steam. 

Artificial intelligence is rapidly transforming human resources operations. From AI-powered recruiting platforms that screen résumés and rank candidates, to onboarding tools that personalize new-hire experiences, performance management systems that predict attrition, identity verification solutions using biometrics, and platforms administering ERISA-governed benefit plans, employers are increasingly relying on third-party vendors whose products are built on or enabled by AI. For labor and employment lawyers, this shift demands a fresh look at how we negotiate and structure vendor agreements.

While most states have not yet enacted AI-specific employment legislation, all employers deploying such technologies should pay careful attention to the terms in their AI vendor agreements. A patchwork of state laws is emerging across the country, federal agencies are actively extending existing frameworks to AI, and even a vendor’s home-state obligations can shape the contractual landscape. This article offers a high-level overview of several key issues practitioners should consider when negotiating agreements with HR technology vendors that use AI.

Define What You’re Buying

Vendor agreements should clearly define service provided under the agreement, capturing not only products explicitly marketed as AI but also any service that includes, incorporates, or is based upon or enabled by artificial intelligence, including generative AI, and agentic AI systems that autonomously act and make decisions in response to prompts and other actions.

Equally important is defining the vendor’s obligation to disclose the nature and extent of its use of AI. Require that the vendor mark or otherwise identify for Customer in writing all deliverables that incorporate or derive from AI, and provide upon request, a full disclosure of the use of AI services in connection with the preparation of any deliverables. Similarly, consider requiring the vendor to notify the employer of a new AI use case, particularly when the current services do not involve AI. Without this transparency, employers cannot assess their own compliance obligations under emerging state and federal laws.

Data Ownership and Restrictions on Training

HR data is among the most sensitive information an organization holds. Vendor agreements should unambiguously establish employer ownership of all content, prompts, inputs, and—critically—all generated output, meaning results or derivative works created by the AI using the employer’s data. Perhaps most important, the agreement should expressly prohibit the vendor from using employer data to train, retrain, or improve its AI models without approval. This is not a theoretical concern—HR data used to train a general-purpose model could expose the employer to claims under data privacy laws and compromise confidential workforce information.

Bias Testing and Anti-Discrimination Protections

The risk of algorithmic bias is the headline issue for AI-driven decision making, particularly in certain areas including health care, housing, financing, and, of course, employment. A growing number of jurisdictions are imposing affirmative obligations on employers who deploy these tools.

New York City’s Automated Employment Decision Tool (AEDT) law, for example, requires employers to ensure an independent bias audit was conducted within the past year before using an AEDT for hiring or promotion decisions, and to publish a summary of the audit results. California’s Civil Rights Department regulations do not have an express bias audit requirement, but recognize that evidence of anti-bias testing or similar proactive efforts to avoid unlawful discrimination is relevant to defending a claim. A similar rule is present in Connecticut’s new AI law.

Because of these regulatory developments, and considering the extent to which organizations rely on vendors for these technologies, employers should negotiate the right to request and receive documentation of the vendor’s bias testing methodology, frequency, and results, and include contractual obligations for the vendor to maintain governance standards and validation procedures.

Cybersecurity Assessment and AI Incident Response

Before onboarding any AI vendor (or, really, any vendor that processes confidential or personal information), employers should conduct a thorough cybersecurity assessment, particularly where the vendor will process biometric data, health information, or other sensitive employee data. The agreement should require the vendor to maintain a written information security program that includes, without limitation, administrative, physical, technical, and organizational safeguards. The agreement also should establish a robust AI incident response protocol, requiring the vendor to notify the employer promptly in writing of any AI incident, and to take corrective action to remediate the issue and prevent recurrence.

Change Management

AI models are not static, they are updated, retrained, and sometimes fundamentally altered. A performance management tool that worked as expected last quarter may behave differently after a model update. Vendor agreements should require advance written notice of material changes to AI services, with enough lead time for the employer to assess potential impacts. The agreement should also grant the employer the right to test and validate changes prior to implementation and require the vendor to maintain appropriate version control, monitoring, and change management procedures throughout the lifecycle of the AI services.

Subcontractors and Third-Party AI Providers

Many vendors do not build their AI models in-house; they integrate third-party AI models or systems, APIs, and/or subprocessors. The agreement should require the vendor to disclose all third-party AI providers, including the nature and scope of their access to employer data. And, liability for those downstream risks to data should be made clear. Preferably, the vendors should remain liable for all acts and omissions of its subcontractors to the same extent as if the acts or omissions were those of the vendor.

Record Retention

Emerging AI laws are creating new recordkeeping obligations. Colorado’s AI Act, for instance, requires deployers to “maintain records to demonstrate compliance” for at least three years after the date of a consequential decision, which may include version identifiers, change logs, and documentation of mitigation changes. Vendor agreements should include record retention obligations that align with these requirements and ensure that the employer has access to the records it needs to demonstrate compliance.

Liability Allocation: Lessons from Colorado and Beyond

Liability allocation is where vendor negotiations often get most contentious. Colorado’s AI Act offers an instructive framework: it provides that liability in discrimination actions involving AI shall be “allocated based on relative fault” between developers and deployers. Colo. Rev. Stat. § 6-1-1707(2). Notably, the Colorado law:

shields a developer from liability when the developer’s covered AI tool is used by the deployer in the manner that was not intended, documented, marketed, advertised, configured, or contracted for by the developer.

In other words, if an employer repurposes an AI tool for a use the vendor never intended or contracted for, the employer may not shift liability to the vendor. For the deployer, managing the use of the tool to align with the services agreement (i.e., governance) will be critical.

This statutory framework underscores the importance of clearly defining the permitted use of AI services in the vendor agreement and having the governance structure to abide by those provisions. A strong indemnification clause should cover claims arising from allegations that the AI service or its output infringes third-party intellectual property rights, privacy and security has been breached, and breaches of AI-specific obligations.

Looking Ahead

The regulatory landscape for AI in employment is evolving rapidly. Connecticut’s new AI law takes effect in October 2027 and will require employers to provide certain disclosures, including the purpose of the technology, the categories of personal data analyzed, and the data sources used. California’s CCPA automated decisionmaking regulations, with a compliance deadline of January 1, 2027, will require pre-use notices, opt-out mechanisms, and risk assessments. 

For employers, the message is clear: the critical role vendors play in AI deployment, together with the combination of multi-state compliance obligations, federal agency guidance, the risk to personal information, and the sheer pace of AI adoption in HR, means that vendor agreements are a foundational component of compliance and risk management. Taking the time to negotiate comprehensive, AI-specific terms today will pay dividends when, not if, the regulatory and litigation landscape catches up.

Senate Bill (SB) 1130, legislation that would establish criminal penalties for certain uses of wearable recording devices, continues to move through the California legislature. I’ve had the honor of discussing this measure with staff of the bill’s sponsor, California State Senator Eloise Gómez Reyes, and anticipate there will be more efforts to enact laws seeking to impose measured responses to the privacy, security, and other challenges posed by the latest generation of AI-enabled wearables.

The Rise of AI-Enabled Smart Glasses

SB 1130 arrives at a critical moment. Modern AI glasses blend high-resolution cameras, always-on microphones, and real-time AI assistants into a hands-free wearable that can capture, analyze, and even transcribe ambient information around the wearer. Unlike traditional recording devices that require deliberate action, AI glasses and similar wearables can passively capture and transcribe conversations throughout the day, creating permanent searchable records of discussions that participants never knew were being documented.

Several institutions and organizations have taken steps to minimize the impact of these devices. For example, in July 2026, New York became the first state to prohibit AI-enabled smart glasses in all state courthouses. Outside the U.S., in August 2026, England and Wales followed suit when His Majesty’s Courts & Tribunals Service (HMCTS) announced that AI glasses will be confiscated from anyone entering its judicial buildings. The UK has also seen its first criminal prosecution involving smart glasses, with a guilty plea for voyeurism at Warrington Magistrates’ Court after a man recorded sex with a woman without her consent using smart glasses.

As we explored in our four-part series on AI glasses, “The Hidden Legal Minefield: Compliance Concerns with AI Smart Glasses,” these devices no doubt raise compliance issues spanning biometrics, two-party consent, workplace surveillance, labor law, data security, and third-party AI processing risks.

What SB 1130 Would Do?

Specifically, SB 1130 would make it a misdemeanor to operate a wearable recording device to capture sound or video of another person in any area within a place of business where that person has a reasonable expectation of privacy, without their explicit consent. Penalties include up to one year in county jail, a fine of up to $1,500 per violation, or both.

The bill also targets the circumvention of recording indicators such as small lights or sounds that signal a device is actively recording. Disabling an indicator would itself be a misdemeanor, and the manufacture, sale, or use of technology primarily designed to disable recording indicators would carry civil penalties of up to $2,500 per violation.

What is a wearable recording device?

SB 1130 defines wearable recording device to mean:

“any device that is designed to be worn on or attached to the body, rather than held by the user, that has the capacity to make sound or video recordings or to transmit sound or video to another device or to the internet.”

This definition likely would capture a broad array of devices beyond AI glasses. However, recent amendments adding “rather than held by the user,” potentially indicate an intention to avoid capturing smartphones, which generally are not designed to be worn or attached to the body (parents of 13-17 years old children may disagree) and typically are held by the user.

What is a place of business?

SB 1130 defines a place of business to mean:

“any physical office or retail establishment in which members of the public receive goods or services from the business.”

What is a reasonable expectation of privacy?

It is not clear at this point and may never be. While walking down a public sidewalk in California does not generally give rise to a reasonable expectation of privacy for visual observation, California law can be more protective than many other states when it comes to audio recording of conversations, targeted or technologically enhanced surveillance, and publication of private facts, even when the underlying events occurred in technically “public” spaces. The context, the nature of the information, and the technology used all matter.

This may be why SB 1130 focuses specifically on places of business where a person has a reasonable expectation of privacy, rather than attempting to regulate all public recording.

Are there provisions which would affect some employees directly?

The bill carves out of the definition of wearable recording devices:

“a headset, two-way radio, or similar device that is operated by an employee during the normal course of their business duties and is provided by their employer for that purpose.”

This description potentially includes positions such as call center representatives and order processors at some fast food restaurants. While providing some relief from its reach, SB 1130 also provides that employees who record sound or video of a customer using such devices must inform the customer that they are being recorded. So, in addition to assessing the implications of more cutting-edge technologies entering the workplace, some employers may need to revisit how their employees use more traditional recording equipment.

What comes next?

SB 1130 has not yet been enacted and must still clear the full legislature and receive the Governor’s signature. But the trajectory is clear: California is moving to close the gap between legacy wiretapping statutes and modern wearable technology and, as noted, it is part of a global trend.

We will continue monitoring SB 1130 as it moves through the legislative process. In the meantime, if you have questions about this bill or related issues, contact a Jackson Lewis attorney to discuss.

AI notetakers illustrate how artificial intelligence is fundamentally changing how people work. For example, there are now “AI Assistants” that snap onto the back of cell phones like a wireless battery pack. Instead of taking detailed notes during a meeting, preparing a summary, and identifying follow-up items, employees can use an AI notetaker to complete these tasks almost instantly. These tools can reduce administrative burdens, improve follow-up, and allow employees to focus on a conversation rather than on taking notes. An assistant, wrapped in a small, inconspicuous package.

Until recently, many AI notetakers were relatively easy to identify. A virtual notetaker, for example, might appear as a participant in a videoconference, which gives the meeting organizer and other attendees an opportunity to approve its use or remove it from the meeting. Organizations could also restrict which applications employees installed on company devices or connected to company systems.

But what if the AI notetaker does not appear on the meeting’s participant list? What if it is already sitting on the conference-room table, or in an employee’s pocket?

  1. What are Portable AI Notetakers?

Portable AI notetakers can record in-person conversations and telephone calls, store recordings while offline, and later synchronize the recordings to generate searchable transcripts, summaries, action items, and other AI-enabled content. Some are small enough to fit in a pocket or attach to a cell phone. As a result, an AI notetaker no longer needs to join a virtual meeting, operate on an organization’s network, or be installed on an organization-issued computer. An employee-controlled device may be able to record a conference-room discussion, interview, telephone call, or informal workplace conversation and later send that information to a third-party AI service without the organization’s knowledge.

While portable AI notetakers may provide meaningful benefits, they can also expose organizations to significant privacy, security, employment, and litigation risks.

  1. What Notice and Consent Issues Can Portable AI Notetakers Create?

Federal and state laws may restrict the recording of telephone calls and other communications. In most states, the consent of one party to the conversation is sufficient. Other states generally require the consent of all parties. The analysis can become particularly complicated when participants are located in different states or countries. Additionally, portable devices expand the settings where the issue may arise, including job interviews, customer calls, workplace investigations, accommodation meetings, and conversations involving patients or visitors.

Even when participants consent to being recorded, that consent alone may not authorize the recording and associated information to be shared with a third-party AI vendor. Participants may not understand that the service will receive their voices and statements and use that information to create a transcript, identify speakers, summarize the discussion, and generate additional records.

  1.  How Can Portable AI Notetakers Circumvent Existing Security Controls?

Many organizations have implemented controls to prevent employees from using unapproved AI tools on their systems. An organization might block AI meeting bots, restrict browser extensions, prevent software installations, or limit access to public AI platforms from company devices.

Portable AI recorders can bypass these controls. An employee may be able to record a workplace conversation without installing software on an organization’s computer or connecting the recorder to its network, then upload the recording through a personal device or account. This is a practice known as “shadow AI”: technology used without organizational approval or oversight. The organization may not know which tools are being used, what information is collected, where it is sent, which third parties process it, or how long it is retained.

  1. What Conversations are at Risk of Exposure?

Workplace conversations frequently include information that should not be disclosed outside the organization. A discussion may concern an employee’s medical condition, a workplace complaint, customer or patient data, business strategy, or trade secrets. A portable recorder may capture incidental information that would never appear in formal meeting notes, such as an employee discussing a family member’s health, a trip to a place of worship, or another personal matter unrelated to the purpose of the meeting.

Outside the workplace, there may also be discussions with customers, auditors, government regulators, legal counsel, or other unsuspecting parties containing sensitive or privileged information. These third parties may have their own confidentiality expectations or legal protections that an undisclosed recording could undermine.

An AI recorder captures what participants say, not just what the organization ultimately determines should be memorialized. Participants may speculate, discuss incomplete information, raise concerns that are later determined to be unfounded, or change their views as a discussion progresses. An AI-generated summary may nevertheless preserve, reorganize, and present those statements as a searchable (and perhaps incomplete) record.

  1. Why Does an AI Vendor’s Privacy and Security Matter?

When an employee uses an unapproved AI notetaker, the organization loses the opportunity to evaluate whether the vendor’s privacy, security, and contractual protections are appropriate for the organization’s intended use. Marketing statements or general compliance representations may not establish which specific AI services are covered, which third parties process the information, where the information is stored, or whether the vendor will accept appropriate contractual obligations.

Before approving an AI notetaker, an organization should understand what the service collects and how that information travels through the vendor’s systems. That review should account not only for the audio recording, but also for transcripts, summaries, prompts, speaker labels, voice data, integrations, and other information generated from the conversation.

  1. What Policies, Approval Processes, and Training Should Organizations Implement?

Organizations should not wait until they discover an unauthorized recording to address these risks. There are several measures organizations can take to help manage the use of portable AI notetakers before sensitive information leaves their control.

  • Adopt a Workplace Recording Policy: Organizations should consider adopting a workplace recording policy that generally prohibits employees from recording workplace conversations without (1) prior organizational approval and (2) the knowledge and express consent of all participants. The policy should apply broadly to audio and video recordings made through electronic devices, including personal devices. It should also identify categories of conversations that may not be recorded even with consent, such as discussions involving trade secrets, personally identifiable information, or privileged communications. However, organizations must be careful to ensure that any recording policy does not restrict employee activity protected by applicable labor and employment laws.
  • Implement an AI Workplace Usage Policy: Organizations should also adopt or update an AI workplace usage policy that addresses AI tools by function, not merely by brand name or software format. The policy should require approval of both the specific tool and the proposed use, prohibit employees from conducting organization business through personal AI accounts or unapproved devices, and make clear that information submitted to an AI service should be treated as a disclosure to a third party. Approved tools and permitted purposes should be identified, while uses involving confidential information, employment decisions, regulated data, or other higher-risk activities should require additional review.
  • Establish an Approval and Vendor-Review Process: An approval process should evaluate the particular tool, the proposed use, the people whose conversations may be recorded, and the categories of information likely to be captured. Approval for routine internal meetings should not automatically authorize recording workplace investigations, accommodation discussions, privileged communications, patient interactions, or customer calls. Organizations should also decide who has authority to approve a tool and who may require recording to stop.
  • Employee Training: While written policies and procedures are critical, they will have limited value if employees do not understand what AI tools are permitted or prohibited. Training should explain the risks of portable AI notetakers, the organization’s approval process, when notice and consent are required, and which conversations may not be recorded. Managers, human resources personnel, and meeting leaders should also know how to respond when they identify an unauthorized recording device or learn that a workplace conversation has been recorded.

***

AI notetakers can provide significant value and convenience. As portable AI notetakers become more common, organizations should ensure that their policies and procedures account for these devices. Organizations that focus only on the AI bot in the virtual waiting room may miss the recorder already sitting on the conference-room table. Policies, diligence, and training should address not only who joins the meeting, but also which devices are listening, where the resulting information goes, and what happens to it next.

If you have heard of CIPA, BIPA, GIPA, or TCPA litigation, you may have an idea of where this post is headed. These acronyms reference federal and state laws that permit a private right of action for certain privacy-related claims, affording successful plaintiffs with statutory remedies. In a recent case, Bartholomew v. Parking Concepts, Inc., an appellate court in California may have opened the door to a new line of similar litigation stemming from a state law regulating automated license plate recognition (“ALPR”) technology.

What is ALPR?

Automated license plate recognition (ALPR) technology uses fixed or mobile cameras paired with computer algorithms to capture license plate images and convert them into searchable, computer-readable data.

Why is ALPR used?

Common uses include locating stolen vehicles, identifying wanted individuals, managing parking access and payment, supporting toll collection, investigating crimes, and monitoring access to secured facilities. For example, Flock, an AI startup that leverages ALPR technology, has been cited for helping to solve crime, albeit not without some controversy.

Is ALPR use becoming more common?

Yes. ALPR systems are often used in the public sector, usually by law enforcement, but private sector use cases have been increasing with parking operators, repossession companies, insurers, HOAs, retail/commercial property owners, and others, as noted in a recent Car and Driver article.

Are there federal requirements for ALPR?

No comprehensive federal ALPR statute exists, although some legislative efforts have been made to limit their use nationally. Regulation is primarily state-by-state, though federal privacy and surveillance law (e.g., Fourth Amendment case law limiting warrantless, prolonged tracking) can inform how ALPR data may be collected, retained, and shared, particularly by government users.

What states regulate ALPR?

At least 16 states have ALPR-specific statutes, including California, Arkansas, Colorado, Florida, Georgia, Maryland, Minnesota, Montana, Nebraska, and New Hampshire, among others. Roughly six of those states restrict private-sector or government use outright, about eight impose data retention limits, and several exempt ALPR data from public records disclosure.

What does California’s ALPR law require?

California’s ALPR statute (Civil Code §§ 1798.90.5–1798.90.55) applies to both government and private-sector “ALPR operators” and “ALPR end-users.” It requires:

  • Reasonable security safeguards (administrative, technical, physical) to protect ALPR information from unauthorized access, use, or disclosure.
  • A written, publicly available usage and privacy policy; if the entity has a website, posted conspicuously on that site.
  • The policy must address: authorized purposes for collecting/using ALPR data; job titles of personnel with authority to use and access the ALPR system; monitoring and audit procedures; rules on sale, sharing, or transfer of data; the custodian responsible for compliance; accuracy safeguards; and data retention/destruction timelines.
  • Public agencies generally may not sell, share, or transfer ALPR data except to another public agency.

Note also that California’s data breach notification law (Cal. Civ. Code Sec. 1798.82) provides that information or data collected through an ALPR system, if breached, could trigger a notification requirement.

Does failing to post an ALPR policy online create liability, even without a data breach?

Potentially yes, at least in California. In the Bartholomew v. Parking Concepts, Inc. case noted above, the court held that a parking garage’s camera system qualified as an ALPR system, and that operating it without a publicly posted usage and privacy policy violates an individual’s statutory “right to know,” which alone is sufficient to allege cognizable harm under Civil Code § 1798.90.54, although not every violation of the law will trigger relief for a plaintiff. The court reasoned:

“In other words, while the ALPR Law does not impose specific restrictions on the use of ALPR information, it grants individuals the right to know which entities are collecting their ALPR data and how it is being used and maintained.  Collecting and maintaining individuals’ ALPR information without implementing and making public the statutorily required policy harms these individuals by violating this right to know…  If an ALPR operator has failed to implement and make public the statutorily required policy establishing authorized uses, it is much more difficult to hold them accountable for unauthorized uses, even though this is an example of a harm-causing violation expressly stated in the ALPR Law.  This further underscores the significance of the publicly available policy to the ALPR Law’s statutory scheme”

What can a plaintiff recover under California’s ALPR law?

Section 1798.90.54 allows an individual “harmed” by a violation to sue any person who knowingly caused the harm, and recover: actual damages (or liquidated damages of at least $2,500 per violation), punitive damages for willful or reckless violations, attorney’s fees and litigation costs, and equitable relief.

Do any other states offer a private right of action or statutory damages like California?

Some do. For example, an Arkansas law (Ark. Code § 12-12-1807) permits a person injured in business, person, or reputation by a violation may sue for actual damages or liquidated damages of $1,000, whichever is greater, plus litigation costs. Nebraska (Neb. Rev. Stat. § 60-3208) allows a private suit for damages that proximately cause injury to business, person, or reputation, but sets no liquidated-damages floor. Washington’s new Driver Privacy Act (SB 6002) creates a civil remedy for an injured person and, for certain government contractors, treats violations as per se unfair/deceptive acts under the state Consumer Protection Act, which carries its own private right of action, treble damages up to $25,000, and attorney’s fees.

What steps should businesses take to comply with the California ALPR law?

  1. Determine whether any camera/software system used (including third-party parking, security, or access-control vendors) meets the statutory definition of an ALPR system.
  2. Draft a written usage and privacy policy covering all statutory elements (purpose, authorized personnel, monitoring/audit process, sharing restrictions, custodian, accuracy measures, retention/destruction schedule).
  3. Post the policy conspicuously on the company website, not merely buried in an internal manual.
  4. Implement reasonable administrative, technical, and physical security safeguards for ALPR data.
  5. Update incident response plans to include ALPR system data.
  6. Restrict data sharing and sale consistent with the policy and applicable law.
  7. Audit vendor contracts (parking operators, repossession firms, security vendors) to confirm their ALPR practices align with the business’s own compliance obligations.
  8. Monitor other states where the business operates, since requirements vary significantly by jurisdiction.

For much of the past two years, discussions regarding generative artificial intelligence (AI) in professional services seems to have focused on lawyers, and perhaps for good reason. Courts have sanctioned attorneys who submitted briefs containing fabricated case citations. In response to these and other mishaps, several state bars issued ethics opinions often applying existing professional obligations to AI such as technological competence, confidentiality, supervision, and billing. Those same themes, however, are increasingly relevant to other professional service providers, such as tax professionals, including employee benefit plan advisors in some cases.

When Do the IRS OPR Guidelines Apply?

The IRS has now made clear that ethics and compliance obligations extend well beyond litigation and the legal profession generally. In June 2026, the Internal Revenue Service’s Office of Professional Responsibility (OPR) issued Introductory Guidelines for Responsible AI Use in Federal Tax Practice (Alert 2026-19), explaining how existing Circular 230 obligations apply when tax practitioners, including attorneys, certified public accountants (CPAs), and enrolled agents, use generative AI. Specifically, the Alert arises out of IRS Circular 230 which governs practice before the IRS and establishes standards of competency, diligence, ethical behavior, and procedures for discipline. Rather than creating AI-specific rules, the Alert applies these existing obligations to the use of AI. It confirms an emerging regulatory principle likely to apply across virtually every licensed profession: AI may assist professional judgment, but it may not replace it.

The work of tax professionals extends well beyond Form 1040 preparation (individual federal tax return preparation). By way of example, professionals working on corporate tax issues, payroll taxes, estate taxes, and the qualified plan rules for employee benefit plans may all need to consider Alert 2026-19 when incorporating AI into their practice.  The Alert offers a roadmap for ethical AI use that such professionals can adapt when performing tax-related services and advising clients.

How Do The Guidelines Address Accuracy?

One of the central themes of the Alert is that AI-generated work must always be independently verified, a principle familiar to every practicing attorney.

When using GAI, practitioners must thoroughly review all AI-created documents and language incorporated into writings before delivery to a client or submission to the IRS. Due diligence requires verifying the accuracy of facts, citations, and calculations produced by AI. Practitioners cannot rely solely on AI; human scrutiny and editing are essential to ensure correctness and compliance with IRS expectations.

In short, AI-generated content should be treated as a draft requiring professional review rather than as a final work product, whether that content is a tax return filed with the IRS, a tax memorandum to the client, or a benefit plan communication addressing tax consequences.

How Do The Guidelines Address Competence?

The IRS also emphasizes that professional competence increasingly requires understanding the technology itself.

Practitioners must understand both the law and the technology used in their representation of clients before the IRS, including AI systems’ operational mechanics, limitations, and risks.

They must understand how AI develops content, recognize the potential for bias or errors, and be able to evaluate whether AI outputs are suitable for use in IRS matters. Lack of technological competence could lead to improper advice or flawed filings.

This mirrors a growing consensus among regulators and professional organizations that technological competence is no longer optional. Professionals need not become computer scientists, but they should understand:

  • when AI is appropriate to use;
  • when independent research or analysis remains necessary;
  • what information AI systems retain or transmit;
  • what risks exist regarding confidentiality and cybersecurity; and
  • how AI outputs should be validated before being relied upon.

Professional competence today increasingly includes the ability to use and supervise the use of AI effectively. The challenge of addressing AI “hallucinations” is not limited to litigation attorneys who fail to identify miscited or nonexistent caselaw. A Bloomberg Tax article, AI Hallucinations in Tax: The Risks and How to Mitigate Them, discusses several examples of AI hallucinations facing tax professionals:

Errors that silently compound across calculations. A single inaccurate assumption can flow through provision calculations, effective tax rate analyses, and disclosures without immediate detection. These errors can then distort financial reporting across entities, jurisdictions, and reporting periods.

Understanding the limitations of the technology is critical for all users and, in particular, for those with professional obligations.

How Do The Guidelines Address Confidentiality and Data Security?

The IRS guidance also underscores another issue rapidly becoming central to AI governance: protection of confidential client information.

Practitioners must strictly handle all client data using only secure, enterprise-approved AI. AI systems should be utilized with robust confidentiality safeguards firmly in place. Willful mishandling of taxpayer information through AI may lead to disciplinary actions under Circular 230.

Whether preparing individual tax returns, payroll taxes, or representing a client in connection with a tax audit of its qualified retirement plan, tax professionals very likely are processing their client’s sensitive confidential information, which may not be limited to personal information. Including such information in AI prompts or documents uploaded to an LLM can present significant risks to the confidentiality of that information. Among the questions organizations should consider are:

  • Has the AI vendor contractually agreed not to use prompts or uploaded information to train its models?
  • Is client data encrypted in transit and at rest?
  • Where is information stored?
  • Does the platform comply with applicable privacy laws and contractual confidentiality obligations?
  • Are employees prohibited from entering confidential client information into consumer AI platforms?

What Role Do Governance, Policies and Procedures Play Under the IRS Guidance?

The IRS makes clear that individual diligence is not enough on its own:

  • Firms must deploy internal policies and procedures for compliance with Circular 230 in the AI space, with coverage that includes:
    • Staff: Comprehensive training on use of AI (the risks, technological and otherwise, and the requirements).
    • Rules applicable internally (within the firm): Established protocols for secure data handling, AI accuracy monitoring, etc.
    • Contracting with external providers: Outsourced or third-party AI tools should be vetted.

Organizations that provide professional services, including law firms, accounting firms, consulting firms, engineering firms, benefit consultants, and healthcare providers should treat AI governance not merely as an IT initiative, but as an enterprise-wide professional responsibility. For example, policies governing approved AI platforms, employee training, confidentiality safeguards, validation procedures, documentation, and client communications are quickly becoming as important as traditional cybersecurity policies.

How Do The Guidelines Address Client Fees?

Perhaps the most interesting portion of the IRS guidance concerns professional billing.

Circular 230 prohibits unconscionable fees, and the OPR goes further, cautioning that billing should reflect any efficiencies AI brings to research, drafting, or analysis.

Practitioners should not only disclose, in general or specific terms as needed, the AI activities performed, but also fairly credit to the client’s account any cost reductions.

For example, if AI reduces an eight-hour task to two hours, firms should consider how that efficiency affects client billing—though the guidance does not address offsetting costs, such as AI licensing, training, and maintenance.

Looking Ahead

The IRS guidance is noteworthy not because it creates new obligations, but because it reinforces a broader reality. Generative AI is changing how professionals perform their work. It is not changing who remains responsible for that work. Whether the practitioner is a lawyer, CPA, enrolled agent, engineer, consultant, or another licensed professional, the emerging regulatory message is remarkably consistent: AI may be an extraordinarily valuable assistant, but it is not the professional of record nor a substitute for sound judgment.