(Editor’s note: This is the second of a three-part series about artificial intelligence in California workers’ compensation.)

Donald R. Barthel

Artificial intelligence isn’t just a law firm problem that stays behind closed doors. When a defense attorney decides to outsource his brain to a robot with the legal judgment of an overconfident first-year law student, the consequences don’t stay in counsel’s office; they land squarely on the claim file.

When defense counsel makes a reckless AI blunder, the file takes a very long, very expensive detour. Algorithms don’t carry malpractice insurance. They don’t have a California Bar license. And they certainly aren’t going to write a check to reimburse a litigation budget.

Instead, the adjuster gets the bill. It paid defense counsel to draft a brief, figure out why the judge is screaming about fake cases, research why sanctions shouldn’t be slapped on the firm and paid to write a groveling letter of apology to the Workers' Compensation Appeals Board explaining why “the computer made me do it.”

Spoiler alert: The “computer error” defense works about as well as telling a judge your dog ate your brief. The WCAB has signaled zero tolerance for unchecked AI output. Under California Labor Code § 5813, judges are handing out notices of intent to issue monetary sanctions of $2,500 or more against attorneys, law firms and claims administrators who file hallucinated pleadings. (For more on this, check out Presiding Judge Robert Rassp’s relatively recent summarization of case law starting on page 16 of this PDF. The sanctions are alarming.)

Now, you might be thinking, “Wait a minute. Applicants' attorneys only get 9% to 18% out of the applicant’s award, or maybe a § 5710 fee for a deposition. How does this cost us extra?”

Prepare to be horrified.

When defense counsel submits a brief containing hallucinated case law or made-up quotes, applicants' counselors don’t just chuckle; they hit the jackpot. Under Labor Code § 5813 and 8 CCR § 10421, a judge can order sanctions and full fee-shifting for bad-faith tactics. Filing a brief with completely fictional law fits that definition perfectly.

Applicants' counsel will immediately file a petition for sanctions. The defense firm, or your claims department, will then be ordered to pay opposing counsel’s hourly rate ($300 to $500) for every single minute they spent hunting down your fake cases, drafting the petition and attending a hearing to tell the judge how incompetent the defense was.

Conclusion

On top of paying opposing counsel to roast your lawyer in open court, you (or your attorney; I’ll let you have a friendly chat about that) get to pay a $2,500 statutory fine directly to the State of California. By letting an unverified AI write a brief, the defense effectively funds both sides of a completely self-inflicted nightmare. Between opposing counsel’s hourly fees, defense time spent fixing the disaster, and court fines, a minor procedural dispute rapidly turns into a $5,000 to $12,000-plus defense liability. Relying on unvetted AI means letting a robot with no bar license write science fiction while your client pays for the publishing rights.

“Open the pod bay doors, HAL.”

If you think judges will let this slide with a stern finger-wagging, wait until you see what happened in Quinteros v. Harbor Distributing.

In our final article, we’ll look at a real-life horror story where a single unverified AI filing triggered state bar investigations, public shaming and a destroyed career.

Donald R. Barthel is a founding partner of Bradford and Barthel and is based in the firm’s Sacramento office. This entry from Bradford & Barthel's blog appears with permission.