We’ve been tracking the burgeoning use of generative AI by patent and trademark lawyers, previously exploring its potential and pitfalls, including the use of AI before the TTAB, a warning about hallucination, and an IP lawyer’s guide to AI. However, we had yet to see the USPTO actually discipline an attorney for the use of AI in practice—until now.
On March 6, 2025, Anthony Matos, was publicly reprimanded by the USPTO, and ordered to complete two CLE courses on the topic of the use of AI. But what is it about the use of AI that caused Mr. Matos to be disciplined? It is hard to say—did he use AI or not?
According to the disciplinary action, Mr. Matos found himself representing the Defendant Trademark Owner in a TTAB matter, having never actually practiced trademark law. In doing so, Mr. Matos filed a trial brief that cited to multiple cases. In their reply, the Petitioner had some words for Mr. Matos, but did not allege any hallucination, but instead suggested that three TTAB cases were allegedly mis-cited. In other words, as the Petitioner put it, “…Respondents are in the habit of citing cases for the other than of what they hold…” We all make mistakes, right?
Not so fast. When the TTAB opinion was released on April 22, 2024, it went further than to allege a mis-citation. The tribunal claimed that there was quite a bit more. It became clear to the TTAB that there may have been some hallucination. As the Order stated:
At oral argument, we asked counsel for Respondent how he had identified these case citations. He responded that he had used the Internet, elaborating that he had performed searches on Google resulting in identification of blog posts and articles discussing these decisions. When we followed up by asking whether he had read the decisions he purported to identify in that manner, he admitted that he had not. He explained that his law practice does not have access to the case reporters in question.
Critically, the TTAB said that not having access to legal research tools is not an excuse—nor is not correcting the citations after the Petitioner brought it to his attention. However, the Order does not appear to have discussed the use of AI by Mr. Matos. The TTAB ultimately struck the pleading. Fortunately for Mr. Matos, the TTAB still held that the Petitioner’s claims were precluded by contractual estoppel, and therefore they did not address the merits of the claims.
Of course, the OED got wind of this and investigated the matter. It appears, based upon the terms of the Final Order, that Mr. Mathos may have eventually admitted to the use of AI, because he was ordered to take CLEs in that area. Most notably, the OED disciplinary action states that the disciplined is:
predicated on submitting a brief to the Trademark Trial and Appeal Board containing inaccurate citations and non-existent quotes to existing case law and citation to non-existent case law. In violation of 37 C.F.R. § 11.18, Mr. Matos failed to read case citations he found using internet searches revealing discussion of the cases on blog posts and articles before incorporating them into his brief and presenting the brief to the TTAB.
Key Takeaways and Ethical Considerations:
- The Matos case highlights the dangers of relying on unverified information, whether generated by AI, or a Google search. Practitioners who represent others before the USPTO have a duty to verify the accuracy of all information presented to the agency, regardless of its source.
- This case also brings to light the importance of access to appropriate legal research tools.
Clearly the use of AI. I can’t believe he was not charged with lying to the tribunal.