AI gave you the answer. But is it the right answer for your legal problem?
AI will give you immediate answers, clear, confident explanations, and a warm, fuzzy feeling. It’s free or close to free, and it’s convenient.
It even feels private.
But GenAI tools may give you bad legal advice. If you’re facing criminal charges in Placer County, CA, or Nevada County, CA, or any county for that matter, relying only on AI chatbots can leave you facing serious consequences.
Does AI ever provide helpful legal information?
AI programs like ChatGPT, Claude, and Gemini are not really “Artificial Intelligence.” They’re more like highly advanced predictive engines that draw on a vast body of online knowledge.
They are useless for anything that requires real judgment, but can be useful for some legal tasks.
Use AI to:
- Understand legal terminology
- Learn the general court process that might govern your case
- Get a high-level overview of potential penalties for specific criminal problems
If you want advice that goes any deeper than that, AI-generated content will not help you.
Here’s what AI can’t do to help you with your legal issue
- It cannot evaluate the specific facts of your case. It may try to do so, but you can’t rely on its accuracy, and what it doesn’t know can hurt you.
- It will not have any specific details about your jurisdiction. The lived reality of a case can change a great deal between counties. Different judges handle things differently. AI does not know this and can’t know this: it’s a matter of boots-on-the-ground experience.
- Discretion plays a big role in our justice system, especially prosecutorial discretion. AI cannot predict how any given prosecutor will react to an attempt to get charges dropped, dismissed, or reduced.
- AI cannot predict long-term consequences.
Why is using AI for your legal issue risky?
AI can result in you:
- Getting overwhelmed—it’s easy to bury yourself in irrelevant information.
- Wasting time—you might spend a lot of time spinning your wheels or missing important deadlines when you’d have been better off calling an attorney instead.
- Coming up with citations that don’t exist or pulling cases that do not have controlling authority…and if you tell your lawyer to try to chase them down, you’ll end up paying more.
- Missing consequences. AI oversimplifies. It may tell you that your case isn’t serious, but it may not be taking into account the fact that you can lose custody of your children, the right to bear firearms, or your professional license. Had you known, you might have hired a lawyer… but you didn’t because you asked AI.
AI is free…but remember, our initial consultation is free too. You can get a lot of information during our consultation, and the cost to defend your case might not be as scary as you think. You won’t know until you ask. If you truly cannot pay, the California courts may appoint a Public Defender to represent you.
Here at the Cohen Defense Group, we’re not anti-AI. It has its place and its purpose. It’s new and getting better every day. But it cannot serve as a substitute for a good criminal defense attorney.
- It cannot stand in the courtroom for you and speak on your behalf.
- It won’t negotiate with the prosecutor.
- It doesn’t have relationships with prosecutors and judges.
- It doesn’t know the little nuances of how a case may play out in Placer County vs. Nevada County.
- It cannot effectively evaluate the specific, unique facts of your case.
- You are not protected when you use AI: your interaction is not covered by attorney-client privilege, which could mean the difference between going to jail and remaining free.
AI should never be asked to provide specific advice about your unique situation, whatever that may be.
AI is good for general education.
Example: “What does mens rea mean?”
AI can answer that.
AI may even be able to answer a general question like, “Is it legal to carry a tonfa or other martial arts weapon in the state of CA?”
Sometimes.
Here’s an example where we’ve seen AI go wrong.
Everyone thinks their case should be dismissed just because the police didn’t read them their Miranda rights. AI may say this is so.
But did you know there are situations where the police do not have to issue Miranda warnings?
We see many disappointed clients because of this one issue.
You may know the Constitution by heart, but what you don’t know by heart, and what AI doesn’t reference well, are the hundreds of Supreme Court cases that create exceptions and loopholes that law enforcement officers use to their advantage every single day.
No. AI is not an attorney and should never be trusted for legal advice.
App maker DoNotPay, for example, is being sued for presenting itself as an “AI lawyer” and giving bad legal advice to clients.
OpenAI is facing a lawsuit for the unauthorized practice of law.
AI legal advice has not been proven to be reliable.
• AI “hallucinates” case law that doesn’t really exist.
• AI cannot be relied upon to ask the right follow-up questions; it only responds to what you put in the chat box.
• AI does not have relationships with judges and prosecutors, which could make a difference in your case.
Yes, but not in the way you might.
Programs like Westlaw and LexisNexis have AI-assisted research tools that flag and color-code cases that might apply to any given search, noting what serves as established (“black letter”) law, what has been overturned by more recent cases, and what might be helpful. These tools provide no legal advice; they are merely legal resources to help lawyers sort through a vast body of research.
We have definitely seen attorneys get in trouble for not double-checking the case citations they got with AI.
You should ask them if and how they are using it to find out.
Yes, our law firm uses AI-powered legal research tools.
These are not chatbots, and they are not used to give legal advice. They are professional tools designed to help attorneys navigate large volumes of legal information more efficiently.
They help us streamline research, identify relevant case law more quickly, and organize complex information.
That is where their role ends.
AI assists Cohen Defense lawyers. It does not make decisions.
AI is used as a starting point, not a final answer.
It does not make decisions or determine strategy. Every result it provides is reviewed, verified, and analyzed by an attorney. We interpret the law and apply it to the specific facts of your case.
AI can surface information, but it cannot determine what matters or how your case should be handled. That requires legal judgment.
Why human oversight is critical for lawyers using AI
Legal work is not just about finding information. It requires understanding nuance, interpreting facts in context, and making strategic decisions.
AI cannot assess credibility, anticipate how a judge or prosecutor will react, or make judgment calls based on experience. Those are the factors that often determine the outcome of a case.
AI doesn’t change our ethical and professional responsibility to you
Attorneys are ethically required to ensure accuracy and protect client confidentiality.
That means we do not rely blindly on AI output. We independently confirm and analyze all information before using it in your case.
If something is wrong, we are responsible, not the software.
The difference between our use of AI as lawyers and the general public’s use of AI
Many people enter personal legal details into AI chatbots and rely on the answers without verification.
That creates real risk. The answers may be wrong, and these platforms are not protected by attorney-client privilege, so what you share may not be confidential and can be used against you.
Our use of AI is controlled. We do not rely on it blindly, and we independently confirm and professionally analyze all information. We do not rely on it to make decisions. Our use of AI helps with efficiency in your case, but it does not replace our legal expertise and judgment.
Where can AI legal advice go wrong in criminal cases?
AI only answers the questions that have been asked.
A lawyer identifies all the risks you might not think to ask AI about at all.
A case isn’t just about the law. It’s about:
- Meeting strict deadlines
- Local filing practices
- Prosecutorial discretion
- Diversion options
- Evidentiary weaknesses
- Collateral consequences beyond the criminal case
- Early decisions that shape the direction of a case
For example, in Tahoe City, a commissioner usually oversees criminal matters, and retired judges may rotate in when needed. In Truckee, a judge hears the cases. Each of these officials makes their decisions very differently. AI will not know how to turn these differences to your advantage.
In Placer County, judges frequently rotate between departments, and retired judges are often brought in to assist with calendars. The court is also increasingly utilizing commissioners in criminal matters. At Placer County Superior Court, this means that the approach to a case can vary significantly depending on who is presiding on a given day. Each judge or commissioner may have different expectations when it comes to negotiation, motion practice, and courtroom procedure. An experienced attorney learns how different judges prefer cases to be presented, when a case may benefit from being set for a hearing versus resolved early, and how to adjust strategy depending on who is on the bench. AI cannot track or apply these constantly shifting dynamics.
In contrast, Sacramento County tends to have more consistency in judicial assignments, with judges typically remaining in the same department. At Sacramento County Superior Court, that consistency means each courtroom develops its own expectations, preferences, and patterns over time. Judges often have specific ways they like motions argued, cases negotiated, and issues presented. Through repeated appearances, attorneys begin to recognize which prosecutors are more collaborative, which cases require a more strategic or persistent approach, and how to tailor arguments to a particular department’s tendencies. AI does not have access to these relationships, patterns, or practical insights.
These are the kinds of details that aren’t written in any statute or case law, but they can significantly impact the outcome of a case.
Example #1 of where AI legal advice can go wrong: AI asks the wrong questions
Our client, “Amy,” was a passenger in a car. The driver was on probation. Amy was not.
Police stopped and searched the car. There were drugs in plain view in the middle console. Both Amy and the driver had access to the drugs.
Amy felt the search was illegal, but she was wrong. The search was legal because the driver was on probation.
Amy asked AI. AI said, “Don’t get out of the car, they can’t ask you to do that.”
But the police absolutely had the right to do that. Amy didn’t tell AI that she was driving with someone on probation. She only told the chatbot about herself, not the driver. The result: bad advice.
Because Amy didn’t get out of the car, she got into deeper trouble than she would have if she had complied with police instructions.
We were able to help Amy, but this is a great example of a case where AI could have led to additional charges rather than helping Amy as she’d hoped.
Example #2 of where AI legal advice can go wrong: AI sends a veteran to the wrong court
Our client, “James”, was a military veteran who was facing domestic violence charges.
AI told James to enter a plea and go through veteran’s treatment court.
If he had followed AI’s advice, he would have entered a plea deal he never needed to enter.
Instead, he called me. I was able to show the District Attorney (DA) that the alleged victim was making everything up. We were able to provide air-tight evidence that she was the initial aggressor. We asked the DA not to ruin this man’s life with a bad case. The DA used discretion to drop the whole thing, all because I was able to provide information the DA didn’t have at the time of charging.
It’s a stark example of how a lawyer can make everything better.
At least the AI did one thing right. It referred James to our office.
Here’s what James had to say.
“Angel Sanchez and Cohen Defense were the attorneys I was looking for. After being arrested, I didn’t know who to turn to. After my service in the military, I barely trust civilians to do right by me. I used AI to analyze court cases similar to mine and had it find me the best attorney who specializes in veterans’ issues. It suggested Cohen Defense Group and the AI was absolutely correct. The awards on their wall speak for themselves when entering their office. From the moment Mr. Cohen answered my text, I felt assured I had the right place. Angel Sanchez was assigned to my case. I felt she was the right attorney, based on her background, and after the conversations I had with her. From the moment she accepted me as her client, I felt at ease, and I felt things could go in the right direction. The case ended up going in the direction I hoped for. I do know it’s because the DA saw I had support from Cohen Defense Group and Angel Sanchez. I keep telling myself ‘if you hired any other attorney than Angel, I don’t think the outcome would have been the same.'”
Note that if you’re a veteran, there are two types of veteran’s courts. There is a veteran’s treatment court and a military diversion court. AI doesn’t seem to understand the difference, in part because not much information is published about it.
We have had other clients tell us that they wanted to do veteran’s treatment court, simply because they are a vet. But in some cases, military diversion is better, because it allows for pre-plea action. A veteran can really harm their interests by following AI advice to enter a plea.
Example #3 of where AI legal advice can go wrong: AI got the statute of limitations wrong
One client used ChatGPT to try to figure out when the statute of limitations was up for a potential misdemeanor battery charge. He hadn’t heard anything, so he assumed nothing had been filed.
He called Cohen Defense Group to expunge the incident.
He was surprised to find out there was a warrant out for his arrest because the case had been filed a few months prior, and he inadvertently missed the court date. It’s possible a notice got lost in the mail or was never sent. Had he not called us, he could have been stopped for a traffic ticket and ended up behind bars due to this warrant.
We were able to call a bail bond agent on his behalf to see what the bail would be so that he could get everything lined up before he turned himself in to clear the warrant. That was all in an initial consult, and it was the consult that kept him out of jail.
Call for your free consultation today.
Example #4 of where AI legal advice can go wrong: it increases your costs
Some clients become so confident in AI research that they end up costing themselves time and money by bringing in case law, statutes, and citations that do not actually exist.
They come in thinking they have a great case and that all the charges should be thrown out.
As a lawyer, I can’t just tell you that you’re wrong on the spot. I’d be violating my own professional duty and ethics.
I have to read every page of the case law to make sure I didn’t miss anything and that I understand what you’re talking about. That’s time you, as a client, end up having to pay for. Some laws take precedence over others. AI can’t seem to understand this. Lawyers know.
Worse than that, we just feel bad when we have to tell you that this thing you’re hopeful about doesn’t apply in your case because of some tiny detail you forgot to share with AI, or because the AI didn’t realize what mattered, or just plain hallucinated.
AI can give you general information about domestic violence charges, but it shouldn’t be your only source of information.
Do:
• Ask AI to explain the charge and possible consequences.
• Ask AI to analyze attorneys in the area who handle domestic violence cases.
Keep in mind that AI only has the information you give it. Its answers might leave out serious consequences such as loss of child custody and loss of firearm rights. AI generally doesn’t discuss collateral damage like this unless you ask about it.
Do not:
• Tell AI all about the facts of your case.
• Use AI to decide whether or not to accept a plea bargain.
• Assume that AI knows everything.
What AI often misses:
• Often, AI won’t tell you about downstream consequences like automatic restraining or no-contact orders.
• It may not give you accurate information about impacts on custody and visitation.
• It may not tell you that getting into parenting classes could save your ability to retain custody of your kids.
• It may not tell you that when a protective order is issued, you have just 24 hours to get rid of firearms and ammunition, and that you’ll have just 48 hours to file paperwork with the court proving you did so.
• It may not tell you how your housing eligibility may be impacted.
Be aware that AI relies on the information you give it. If you don’t have all the details, you cannot trust AI’s responses. You may not even learn what charges you are facing until 1-2 days before the arraignment. Without specific details like that, AI cannot give you accurate answers about your case. Ideally, you should talk to a lawyer before charges are filed, because:
• A lawyer may be able to keep charges from getting filed at all.
• A lawyer can help you engage in mitigation efforts, like taking anger management classes or gathering helpful evidence.
• Your lawyer can get a head start on tracking down records and reports that will help your case.
• A lawyer can open communications with the DA. An early conversation between your lawyer and the DA can change your trajectory for the better if you move fast enough.
Early legal strategy matters! Early decisions impact the overall direction of every case.
Early decisions include:
• Do you speak to law enforcement at all? (Not without a lawyer present, we hope).
• What do you say? (Nothing your attorney doesn’t tell you to say.)
• Are you allowed to speak with the protected party? Protection orders can be confusing.
• Whether to seek an approved method of communication with the protected party.
• Whether we may be able to push for a peaceful contact order instead of a no contact order.
• How criminal and family court matters may be coordinated and shared between your criminal attorney and your family attorney.
When we meet with you, we always ask:
• What do you have to lose?
• What do you want to protect?
These are the most important questions we can ask. They help guide our strategy for everything else that follows. Even if you tell AI what you stand to lose and what you have to protect, it may miss important actions you could be taking.
AI can provide certain information:
• How to get help with the DUI process.
• Information about punishments and penalties.
• Information about maximum penalties.
Many DUI cases get dealt with pre-filing or out of court.
AI cannot ask the District Attorney (DA) to drop or reduce charges. We can do that because we have a working relationship with the DAs who are making those decisions. They know we cover all our bases and will usually pay attention if we say that our case is strong. We might be able to get a DUI reduced to a “wet reckless,” which means it doesn’t count as a first-time DUI. You may be able to keep your professional license.
AI cannot help you find driver’s license reinstatement classes that the DMV will actually accept. It may try, but if it’s wrong, your license stays suspended.
AI also doesn’t know what you don’t know.
• It might not tell you to read important paperwork.
• It doesn’t explain deadlines.
• It doesn’t tell you how to avoid getting punished twice, once by the court and once by the DMV.
• It can’t tell you how to handle an APS hearing correctly. This is the administrative hearing that determines if your driver’s license is suspended…or not.
• It can’t help you negotiate house arrest instead of jail time.
• It likely won’t tell you about long-term insurance consequences.
AI is only as good as the questions you ask it. If you don’t know what to ask, you could miss vital information.
AI likely cannot tell you if you qualify for expungement. Eligibility for expungement is complicated. AI usually misses the nuance. Here are two examples of where AI went wrong.
Example #1 of where AI was wrong about expungement
ChatGPT had already told client #1 that he was eligible for expungement. It even spat out a few statutes that looked good.
He called for a consult, thinking this was a done deal.
It was painful to tell him he couldn’t qualify. He had been convicted under a sub-statute of the law. AI didn’t understand what made him ineligible, even though he provided the chatbot with the correct information.
Example #2 of where AI went wrong about expungement and getting firearms back
Client #2 saved up for years to hire a lawyer to get his firearms back. He finally had the money.
He asked AI what he should do.
AI told him to do a Certificate of Rehab, and he’d be good. AI was wrong.
The client’s charge didn’t qualify for expungement. He got his hopes up and took unnecessary steps, only to be disappointed in the end.
Unfortunately, AI may miss licensing concerns. It tends to focus on criminal penalties, which can be a huge mistake. You’ve invested a lot in your career… don’t lose it by turning to the wrong advisor to solve your problems.
Criminal charges impact many professional licenses—almost any type that requires you to go in front of a board of ethics:
• Attorneys
• Real estate agents
• Doctors
• Nurses
Any “crime of moral turpitude” can cause issues, even misdemeanors.
Crimes of moral turpitude include:
• Theft
• Perjury
• Burglary
• Fraud
• Deceit
• False statements
• Domestic violence
• Battery
• Assault
• All felonies
While drugs are not considered crimes of moral turpitude, they can still carry consequences: you may be required to go to treatment just to keep your license.
AI can only help with professional licensing concerns in a small way. The best AI can do, safely, is give you a starting point for what to gather so you can show that you’re engaging in a mitigation process. This can help you with professional boards.
For example, if you have a domestic violence charge, then taking anger management classes can help you retain your license, because it shows you’re trying to rectify the problem. There are no guarantees! But it can help.
Nevertheless, mitigation is no substitute for getting an attorney to argue your case on your behalf in front of your licensing board.
While we do not do this kind of work, we can refer you to attorneys who do. We can also put together mitigation packets for that attorney so they’re fully prepared. AI will never do this for you.
Where can AI go wrong about professional licensing issues in your criminal case?
Many professional licenses come with reporting requirements. If you fail to let the licensing board know about the charges you’re facing in a timely fashion, you can end up being penalized just for failing to report!
You need an attorney to research these requirements for you, because every board is different, and so are the crimes you have to report.
For example, some boards don’t require you to report being charged with a DUI, but if that DUI also came with a hit-and-run charge, they absolutely do.
AI will not always know what you do or do not need to report.
One small decision can lead to many problems down the line.
Example of AI being wrong about professional licensing: a nurse’s career at risk
One of our clients was a nurse who was in danger of losing his license. He was charged with public exposure for gratification.
Our office was able to bargain with the District Attorney (DA), and the charge was reduced to a disorderly conduct charge. We then helped our client enroll in mitigation measures, softening the blow and giving him a fighting chance in front of the nursing board. We shared all of this with his licensing defense attorney, who was able to save the client’s license.
Imagine if the nurse hadn’t come to us and had gone to AI instead. He may have been convicted of the more serious charge and could have been fired from his job. This charge would have come up again and again on background checks. He would’ve lost everything. Instead, we negotiated a favorable resolution, and even got the plea withdrawn, an early dismissal, and had the case sealed, which helped tremendously during his licensing board hearing. He retained his license and returned to work a little wiser for the wear.
No. In general, we would advise you to avoid putting any personal information or confidential communications into an AI chatbot. General information is often safer, but be careful because there is no guarantee of confidentiality.
Why? Because AI platforms are not protected by attorney-client privilege.
Do you know what Anthropic, OpenAI, and Google do with all your data?
Neither do we.
Do you know how secure that data is?
Neither do we.
Attorney-client privilege ensures that whatever you say or write to your attorney remains private. It cannot be used against you in a court of law.
This is so incredibly important because your lawyer needs to know all the facts of your case to effectively represent you.
There are ways this privilege can be lost:
• Sharing the conversation you had with your attorney with a third party
• Discussing your defense strategy with anyone other than your attorney
• Sharing confidential details in a discoverable format like an AI chatbot
Once you type any detail about your case into any generative AI chatbot or LLM program, that conversation is stored. Someone may review it. Someone may subpoena it and ask witnesses about it on the stand. Admissions, inconsistent statements, and other details may all be used against you in a court of law.
No attorney-client privilege exists with AI chatbots because AIs are not attorneys.
That means all your chat transcripts may be used against you in a court of law. They could become the evidence that convicts you.
Even if you think you’re being careful with the details you provide, you may still let something slip that hands the prosecution a win.
You should treat AI like a shared file: a Google Doc, rather than a locked and passworded confidential file.
You can also think of it like Googling your symptoms versus talking to your doctor. We’ve all gone down the WebMD rabbit hole. But when it actually matters, you go to a doctor, not just because the conversation is private and protected by HIPAA, but because they have the training and experience to properly diagnose and treat your issue.
Legal issues are no different. Sharing details with a chatbot or even a friend offers no protection and no real strategy. Speaking with your attorney gives you both confidentiality and informed guidance. When you’re discussing conduct that could lead to criminal consequences, privacy isn’t optional — it’s critical.
Things you should never share with AI about your criminal case
NEVER tell AI the following:
- The specific facts of your case. It won’t get you better advice, and it means the prosecutor can use those facts against you.
- Any admission of wrongdoing.
- Any description of your lawyer’s legal strategy—AI can’t evaluate that for you or serve as any kind of a second opinion.
- Any information from your attorney.
- Any details: names, dates, identifying details.
- Any documents, such as police reports.
AI is not your friend. AI is not your lawyer. And unlike lawyers, AI programs are not bound by ethical and professional responsibility to ensure accuracy and protect confidentiality.
You will be tempted. You will want reassurance throughout your case. The chatbot is right there at 3 in the morning. Please do everything in your power to resist the urge. Doing so could be the difference between years in jail and a short diversion program that leaves you free and your record cleared.
Does AI know how much a lawyer costs?
No. If AI told you a number, it’s taking a guess.
Legal prices are rarely set. They are based on:
- The lawyer’s knowledge and experience
- The complexity of the case
- The type of billing the lawyer uses
- Whether you’re calling pre-charge
- Whether a trial is likely
If you want accurate pricing, consider calling attorneys and getting quotes.
AI only knows what is published online. Few attorneys publish their rates online, and every case is unique, so AI cannot give you a valid answer.
AI can’t predict what will really happen in your criminal case
AI may be a predictive engine, but it predicts text, not case outcomes.
- You may forget to tell AI vital details.
- AI may not know what those details mean.
- AI does not ask good follow-up questions.
- AI forgets about collateral damage like the loss of professional licenses.
- AI does not know the local District Attorney (DA).
- AI does not know the judge.
- AI does not have your attorney’s negotiation skills.
Even a good attorney can’t predict your case with 100% certainty. Don’t rely on AI to do it for you.
AI cannot account for things that can make or break criminal cases
For example:
- Local filing practices
- Diversion options
- Informal negotiation opportunities
- Evidentiary weaknesses that may influence charging decisions
- The strategic value of early intervention
AI can’t represent you pre-charge
If you involve attorneys early, we can:
- Present mitigating evidence
- Influence filing decisions
- Potentially reduce charges
- Sometimes secure diversion before formal charging
AI cannot negotiate, intervene, or assess local discretionary patterns.
AI can’t help you avoid charges before they are filed
Attorneys who come into the case before charges are filed (pre-charging) can take action to avoid filing or to get lesser charges filed.
AI can’t help with this.
Pre-charge work requires investigation, evidence gathering, and conversations with local district attorneys.
AI told me to wait to see if charges are filed – should I?
Absolutely not, even if you think AI is telling you some pretty convincing things.
Always, always contact a lawyer right away if you want the best possible outcome for your situation!
AI can inform you, but it cannot defend you
AI can only provide general information.
It cannot:
Manage critical deadlines
- Evaluate local filing practices
- Predict discretionary decisions
- Identify overlapping collateral consequences
- Intervene strategically on a client’s behalf
- Represent you pre-charging
In time-sensitive or high-impact cases, legal guidance can change outcomes.
A free consultation with an attorney will give you more thorough and accurate information than an AI search, and may even be all that you need.
Before you fire up that chatbot, why not schedule your free consultation today?







