On January 1, 2027, California’s Senate Bill 574 will become law, making significant changes to the legal tech landscape and impacting legal professionals using generative artificial intelligence (AI) in their practice. SB 574 makes changes to various California statutes: it adds Business and Professions Code section 6068.1 and amends Section 6173, and it also amends Code of Civil Procedure sections 128.7 and 1281.1 as well as adds Section 180. These changes have a real impact on attorneys, arbitrators, and judges alike.
First, attorneys are charged with not delegating the practice of law to AI. Second, they must verify all AI outputs they receive, including cases and statutes, and make sure they are accurate before using them or correct them. Basically, SB 574 is saying that the “tail can’t wag the dog.” Although AI can provide confident-sounding answers to complex legal questions in seconds, attorneys have to maintain their judgment and not become a mere pass-through for AI. Indeed, fake legal citations have appeared at least 2,200 times in legal briefs to date.
Not surprisingly, SB 574 officially makes violating the above rules sanctionable conduct under Code of Civil Procedure section 128.7.
Third, as part and parcel of not blindly relying on AI outputs, attorneys must “not enter confidential, personal identifying, and other nonpublic information” into AI unless the system is “restricted to the attorney and persons authorized by the attorney under obligations to protect the confidentiality of the information.” SB 574 defines “personal identifying information” to include driver’s license numbers, birthdates, Social Security numbers, National Crime Information and Criminal Identification and Information numbers, addresses and phone numbers for parties, victims, witnesses, and court personnel, medical or psychiatric information, financial information, account numbers, and “[a]ny other content sealed by court order or deemed confidential by court rule or statute.” In other words, attorneys are tasked with understanding the level of security offered by the AI tools they utilize in practice, including the terms and whether the data entered is protected.
Finally, Section 6068.1 requires attorneys to disclose to courts that they used AI to create a filing and asks them to “consider whether to disclose the use of generative artificial intelligence if it is used to create content provided to the public.” Essentially, attorneys have to come up with a clear way of stating in each of their court filings that they used AI. Perhaps a footnote or a sentence at the beginning of the document. And if they are using AI to, for example, draft blog or social media posts, attorneys should “consider” disclosing that fact.
Next, SB 574 provides new rules for arbitrators. Much like attorneys, arbitrators cannot simply delegate their decision-making to AI. What this would look like is an arbitrator uploading all of the case documents into an AI tool, asking it to draft a decision, and then the arbitrator just adopting the AI decision without consideration. Arbitrators cannot rely on AI outputs from outside of the record without disclosing it to the parties and allowing them to comment. One can imagine how parties could lose faith in the arbitration process if the ultimate decision were to depend on some hearsay an AI tool relayed to the arbitrator, as opposed to the verified evidence submitted by the parties. [It should also be noted that SB 574 amends Business and Professions Code section 6173 regarding the right of access to State Bar complaints against alternative dispute resolution providers, but this does not appear to be relevant to AI.]
Last but not least, SB 574 wades into the waters of judicial usage of AI. But rather than expressly provide new rules as it does for attorneys and arbitrators, it simply adds Code of Civil Procedure section 180, which reads: “The Judicial Council shall publicly revisit, and revise as necessary, Standard 10.80 of the California Standards of Judicial Administration to incorporate any necessary changes to reflect the further development of generative artificial intelligence.” Standard 10.80 is the existing AI standard that applies to judges in California. Therefore, we will have to wait and see what adjustments the Judicial Council will make to Standard 10.80 in light of the rapidly evolving world of AI. In fact, judges are no more immune from making mistakes with AI than attorneys as many have also included fake AI law in their rulings.
There is no doubt that states are trying to move fast to pass legislation regulating AI. However, California appears to be the first one to actually pass legislation regulating AI usage by legal professionals. While it is expected that many other states will follow suit, their versions of AI regulation could come in different forms, such as state bar rules, ethics opinions, court rules, local rules, or even courtroom rules.
Therefore, attorneys are truly facing a patchwork of AI rules and regulations right now, especially those who engage in multistate practice. Until a greater level of uniformity is achieved, the only thing we can do is stay on top of applicable AI rules, ranging from state-level to specific judge issued courtroom rules.
