M.D. Pa.: Standing Order on Use of Generative AI (Magistrate Judge Phillip J. Caraballo…
Magistrate Judge Phillip J. Caraballo · U.S. District Court for the Middle District of Pennsylvania
Verified September 28, 2026
- Citation
- Standing Order on Use of Generative AI (Magistrate Judge Phillip J. Caraballo, M.D. Pa.)
- Order date
- January 25, 2025
Summary
Current text: 'All parties and counsel are cautioned to verify any facts or legal citations prepared through generative AI, and to safeguard against the disclosure of confidential information.'
What does the order require?
- Current text: 'All parties and counsel are cautioned to verify any facts or legal citations prepared through generative AI, and to safeguard against the disclosure of confidential information.'
- 'The failure to comply with those practices may result in sanctions.'
- 'All parties and counsel that use Generative AI are directed to review the conclusions on pages 15 and 16' of the Joint Formal Opinion of the Pennsylvania Bar Association and Philadelphia Bar Association on the use of AI 'and be mindful of their ethical and professional obligations before this Court.'
- Earlier text (in place by March 2025): any party using a generative AI tool in preparing a document filed before Judge Caraballo had to include a Certificate of Use of Generative AI disclosing the specific tool, the portions prepared by AI, and a certification that a person checked the accuracy of any AI-generated portion, including all citations and legal authority. The current page no longer requires the certificate.
Practice areas: federal civil
What the chambers rules say now
Magistrate Judge Phillip J. Caraballo’s chambers page has a “Use of Generative AI” section. It notes the risks of generative AI “(including, but not limited to, ChatGPT, Claude, and Grok)”: that the tool “might generate legally or factually incorrect information,” “might create unsupported or nonexistent legal citations,” or “may disclose confidential information in the public sphere.” It then provides:
“All parties and counsel are cautioned to verify any facts or legal citations prepared through generative AI, and to safeguard against the disclosure of confidential information.
The failure to comply with those practices may result in sanctions. Further, all parties and counsel that use Generative AI are directed to review the conclusions on pages 15 and 16 of the Joint Formal Opinion of the Pennsylvania Bar Association and Philadelphia Bar Association regarding the use of Artificial Intelligence and be mindful of their ethical and professional obligations before this Court.”
No certificate or disclosure of AI use is required.
What it replaced
In the version captured on March 4, 2025, the same section required “any party, whether appearing pro se or through counsel, who utilizes any generative AI tool in the preparation of any document to be filed in any matter pending before Judge Caraballo” to “include with the document a Certificate of Use of Generative AI in which the party must disclose and certify: The specific AI tool that was used; The portions of the filing prepared by the AI program; and That a person has checked the accuracy of any portion of the document generated by AI, including all citations and legal authority.” It added: “Failure to comply with this Order may result in sanctions.” Filings made while that version was in place were subject to the certificate requirement.
Primary source
Magistrate Judge Caraballo’s chambers page, pamd.uscourts.gov; earlier text: Internet Archive capture of 2025-03-04
How this order changed
Dated record of this order's own lifecycle. Where it was amended or withdrawn, the badge shows its standing after that step.
- activeIn place by this date (Internet Archive capture): the chambers page's 'Use of Generative AI' section requires a Certificate of Use of Generative AI (tool, AI-prepared portions, accuracy check) with any filing prepared with generative AI. source
- amendedAs read on this date, the section no longer requires the certificate: it cautions parties and counsel to verify AI-prepared facts and citations and protect confidential information, on pain of possible sanctions. The change came after the March 4, 2025 Internet Archive capture; its date is not established.