When the news hit in late September that the cybercriminal group ShinyHunters had breached an FBI recruitment portal – exposing the sensitive information of thousands of current and former employees – it sent shockwaves around the agency and throughout the wider national security community.
While the story received a tremendous of amount of media attention, attacks on law firms are unlikely to get the same coverage, despite the frequency of such incidents and the potential threat they pose to clients.
According to the 2026 Law Firm Technology Survey by cloud and managed IT service provider the Tabush Group, the number of cyberattacks has more than doubled since 2025.
The survey, which was based on responses from over 230 law firm leaders, showed incidents rose from 6% in 2025 to 13% this year. At the same time, 91% of respondents expressed confidence in their firms’ cybersecurity measures.
The damages are also rising.
IBM’s 2026 Cost of a Data Breach Report found the average global incident totaled $4.99 million, up 12% from last year.
It also noted a 56% increase in AI-driven attacks, which average $6 million apiece, with professional services organizations, including law firms, a prime target.
This year, Weil, Gotshal & Manges was the victim of a sizable data extortion incident and Fox Rothchild sustained a data breach, which led to the filing of a class action lawsuit.
In Los Angeles’ own back yard, downtown-based Buchalter was also recently hit, reporting an incident involving their client, Arrowhead Regional Medical Center, in which personal and health-related information was accessed by an unauthorized third party.
In addition to financial costs, breaches can compromise a law firm’s reputation and expose it to litigation, regulatory penalties, and professional-disciplinary actions.
New, growing threat
While law firms have always faced attacks from hackers, cybercriminals are stepping up their game in the age of AI, using the technology to automate phishing scams and create increasingly convincing impersonations of institutions, company leaders and even clients.
A single breach can expose privileged attorney-client information, financial data, intellectual property, regulatory filings and litigation strategies, providing numerous ways for hackers to cash in.
For firms that rely on outdated cybersecurity protocols, the risk is even greater.
“The legal profession is entrusted with a lot of confidential information,” said Christine Lofgren, who teaches Generative AI for Lawyers at Southwestern Law School in Koreatown. “We as lawyers might be holding privileged and confidential client information on a deal or litigation matter or be responsible for client money held in trust accounts.
“This makes attorneys and law firms fruitful targets, and AI is now making it harder to keep that information safe,” added Lofgren, who is also director of the technology law and entrepreneurship concentration.
Protecting the bottom line
Los Angeles County Bar Association Privacy and Cybersecurity Section First Vice Chair Christina Tusan said AI’s ability to gather information at record speeds and create realistic audio and video deepfakes has made gaining access to all types of data easier.
“Wire transfer requests are particularly vulnerable,” said Tusan, founder of Pasadena-based Tusan Law and a former senior trial attorney for the Federal Trade Commission. “If a hacker gets into the email of a settlement administrator, for example, and issues a request it may seem quite authentic.
“Vendor invoices can now be made to look very realistic,” added Tusan. “And the wealth of information available on the internet makes it easier to target those at the top in spear phishing scams.”
Tusan said the new environment makes it imperative to verify any financial transaction at each step of the process, and most importantly, to ensure that the account number where the money is supposed to be sent matches the actual account number of the client or vendor.
“With respect to wire transfers in particular, most people don’t know that if the addressee and wire number don’t match, banks will often still let the wire go through,” said Tusan. “I have seen cases where a business’ email was compromised and a wire request was sent out to the name of a legitimate vendor, but the account number was changed.
“Once the money is sent out as a wire to a scammer,” she added, “it’s very difficult to get it back.”
Experts urge organizations to confirm all financial transfers by phone, if possible.
“I typically tell both sides of a transaction that they should send a secure link related to the transfer and then get on the phone directly, at numbers they know belong to each of them, and verify the transaction before sending out funds,” said Tusan.
Lofgren said it used to be much easier to spot phishing emails due to grammar mistakes or poorly replicated logos, but with AI, these emails can be quite convincing.
“It almost happened to me,” said Lofgren. “I was working on a transaction and expecting a Docusign electronic signature request from someone. A fake Docusign request came in the same day, and it looked pretty real. The logo and text looked genuine. The timing was right, and there were no typos or other red flags that would have alerted me.”
Tusan has also been targeted by phishing scams, receiving an email purporting to be from the City of Pasadena and another from the State of California that appeared to offer a contract opportunity.
In both cases, she recognized them for what they were and informed the agencies.
She said it’s very important for everyone to block and report any phishing attempts or unsolicited calls. Given the increasing threats, Tusan said all firms, regardless of size, need to step up their cybersecurity.
“I’ve seen a case where the hackers got into the CEO’s email and sent an email to staff while she was on an international flight, something they knew because they also had gained access to her calendar,” said Tusan.
“They sent an email directing staff to make a large refund to a particular customer, but using the scammer’s account information,” said Tusan. “I’ve even seen hackers target interns. With all the information online, it’s easy for hackers to pretend to be a trusted person that staff members want to please.”
In addition to emails, Lofgren said hackers can utilize phony QR codes or links to gain entry to firm systems through less-protected personal devices like mobile phones.
“Sellers send codes offering discounts all the time, but they are not always real,” said Lofgren. “Password resets are also another common tactic. All it takes is for one person to use the code or click the link to let a hacker in.
“A law firm can have a highly secure system, but it can all be undone with a simple scan or click,” said Lofgren.
Since not everyone is aware of all the tactics, Lofgren said it’s key for firms to train all their employees on the latest scams.
Threat from within
Threats are not limited to external phishing scams.
As firms increasingly turn to generative AI to reduce mundane tasks such as administrative duties or to streamline the legal research process, experts say the technology itself can provide a point of entry.
“Before adopting an AI product, firms should understand what data the application can access, whether information is retained or used for training, what security protections are available and whether the product was actually designed for legal use,” said Tusan.
Southwestern Law School Associate Professor Tiffany Li, who teaches privacy and technology law, said firms must ensure that only approved tools are utilized by their employees.
“Many platforms offer public, commercially available applications as well as enterprise versions that may feature additional privacy guardrails that limit data visibility to within the firm,” said Li. “However, even if a more secure application is adopted by the firm, a hacker can still gain access to protected information if an employee uses a public, commercial version at home.”
Even more concerning, said Li, if an attorney inputs confidential client information into a generative AI application, the data can lose confidentiality and privilege protections.
“There have been a few cases of attorneys and even clients asking questions using a public AI tool in which the questions then became part of discovery,” said Li. “AI companies do not have a legal duty to keep information confidential, and much like with social media, once the information is out there, it’s too late.”
To help prevent such issues, firms are advised to develop strict policies regarding AI, clearly defining the circumstances in which the tools may be used and banning personal platforms.
Tusan said restricting access to sensitive data is also key.
“Network segmentation can limit the damage if an attacker gains access to one part of a firm’s system,” said Tusan.
“Not every employee needs access to every database,” Tusan continued, adding that information that is no longer needed should be deleted rather than retained indefinitely. “Hackers can’t steal what you don’t have.”
Multi-step assessments
At the Am Law 100 firm Blank Rome, Los Angeles and Orange County partner Sharon Klein said assessing an AI product is a multi-step process.
Klein, co-chair of the privacy, security and data protection practice and firm vice chair of artificial intelligence, said businesses need to understand the specific risk profile of each tool before it’s deployed.
She said Blank Rome conducts security assessments before implementing AI products, including governance controls, auditing capabilities, role-based data access controls, and other features necessary for responsible AI use.
The tools are usually kept in the pilot stage for several months to further assess their risks.
“Our IT people carefully evaluate each program, and a group of attorneys test it out to see if it checks all of our boxes before it’s fully implemented,” she said.
Employees also undergo formal training, Klein said.
“It’s interesting that a lot is reported about AI advancing rapidly,” Klein said. “But what is also advancing rapidly are additional pieces of technology that reduce risk.”
Given the changes, Klein said the guidance has shifted from never putting confidential information into a prompt to what tools, if operated correctly, can allow the firm to do so safely.
“We use AI in a closed environment in which questions are processed within the firm’s own systems rather than being sent into a public model,” Klein said.
Blank Rome utilizes AI to assist with things such as litigation discovery and transactional due diligence, as well as for administrative tasks and training purposes to help improve writing skills.
“It can help generate ideas and tighten writing, but it must never be used to replace legal judgment,” Klein said. “It is simply another set of eyes that can provide a start for human analysis. We must keep in mind that it can also produce hallucinations, including fake citations and cases that can lead to sanctions if they make it into court.”
Lawyers’ duties in AI age
The American Bar Association’s Model Rule 1.1, Comment 8, explains that attorneys must keep up with “changes in the law and its practice, including the benefits and risks associated with relevant technology…”
More recently, Formal Opinion 512, issued in 2024 by the Standing Committee on Ethics and Professional Responsibility, clarifies that the use of generative AI does not create an exception to an attorney’s ethical duties. In other words, they must understand the tools and the potential for errors or hallucinations like fake citations.
The California State Bar’s 2026 Practical Guidance for the Use of Generative AI specifically states, “The duty of competence requires more than the mere detection and elimination of false or fabricated AI-generated results (i.e., hallucinations). Competence, at its core, requires that the lawyer (and not the technology) retains control over strategic decision-making and the ultimate exercise of professional judgment in the representation…As a general matter, a lawyer must not input any confidential information of the client into a generative AI solution that may present material risks to confidentiality or security, absent informed client consent (see rule 1.0.1(e)) as to the underlying risks.”
Diane Cafferata, founder of the Cafferata Litigation Firm in Pasadena, said how this guidance is applied will change as the technology evolves. As a member of the California Lawyers Association’s AI task force, she co-authored the “Report on the Practice of Law,” which was unveiled in 2024.
“One of our conclusions was that the existing professional duties provide a strong framework for addressing AI,” said Cafferata. “The fundamental duties of the practitioner do not change; but every tool the lawyer uses – and AI is no exception – must be carefully evaluated in terms of its function, its limitations and its vulnerabilities.
“Lawyers and law firms remain responsible for complying with those duties,” said Cafferata.
“That requires understanding these tools well enough to ensure to use them appropriately and recognize where their limitations create risk.”
For example, Cafferata said lawyers have always had a duty to protect client confidentiality.
“In 2024, applying that duty to AI meant focusing largely on what happened to confidential information a lawyer provided to an AI system – whether the information was retained, used for training, or disclosed to others,” she said. “As AI becomes more autonomous or ‘agentic,’ that same duty requires lawyers to think more broadly – not only about what information the system can access, but what actions it can take based on that information.”
Leave the lawyering to
the attorneys
Lofgren said she tells her students to view AI as a junior colleague: useful, but in need of supervision.
“This is particularly important when AI is used for legal research,” said Lofgren. “A lawyer who already understands the relevant law may be able to use AI to quickly identify cases, summarize an issue or generate potential arguments and then independently verify the results. “But a less experienced lawyer who cannot recognize an erroneous answer may have difficulty distinguishing a useful result from a convincing hallucination.”
Just because it’s in LexisNexis doesn’t mean the case is real, Tusan said.
“You have to click on the link and read the case,” said Tusan. “It could be an actual pleading that was filed by another attorney but contains a fake AI case, or a decision discussing issues with that fake case that you may unintentionally cite as if it were a real case.”
Cafferata said AI offers tremendous potential, if properly used.
“It can reduce the time lawyers spend processing information and free them to spend more time on what lawyers do best – exercising judgment, developing strategy and advocating for their clients,” said Cafferata.
