Blog

PDF metadata and court filings: what leaves with the file

A filed PDF carries the drafter name, the firm name and the timestamps, in a place redaction never touches. What the rules say, and what to check.

By Harish6 min read

A PDF you file or serve carries a second document inside it: the name on the machine that made it, the software, the minute drafting started and the minute it finished, and often the name of the firm or the client. None of that appears on the page. None of it is touched by a redaction box. Strip it before the file leaves your control, because once it is on a public docket you cannot take it back.

Two obligations, and only one of them is written down

Most filing guidance covers what is visible on the page. Federal practice has a specific rule for it. Rule 5.2 of the Federal Rules of Civil Procedure says a filing that contains a social security number, a taxpayer identification number, a birth date, the name of a known minor, or a financial account number may include only the last four digits of the number, the year of birth, and the minor’s initials. That is a rule about ink.

Nothing in Rule 5.2 says anything about metadata. Metadata is governed by the general duty of confidentiality instead. ABA Model Rule 1.6(c) says a lawyer “shall make reasonable efforts to prevent the inadvertent or unauthorized disclosure of, or unauthorized access to, information relating to the representation of a client.” That sentence was added in 2012 and it is deliberately open ended, which means the standard moves as the tooling gets easier. Comment 8 to Model Rule 1.1 pushes in the same direction: a lawyer should keep abreast of “the benefits and risks associated with relevant technology.”

So there is no rule that names PDF metadata. There is a duty that plainly covers it.

Why the sender carries the risk

In 2006 the ABA Standing Committee on Ethics and Professional Responsibility issued Formal Opinion 06-442, on the review and use of metadata. It concluded that the Model Rules “do not contain any specific prohibition against a lawyer’s reviewing and using embedded information in electronic documents.” It also encouraged lawyers sending electronic documents to remove metadata from them first.

Read those two findings together and the practical position is clear. The receiving lawyer is generally permitted to look. The sending lawyer is expected to have cleaned the file. State bars have split on the first half of that: the Alabama Disciplinary Commission took the opposite view, treating metadata mining as a deliberate attempt to obtain privileged information, and the North Carolina State Bar reached its own position in 2009 Formal Ethics Opinion 1. They have not split on the second half. Nobody anywhere has said it is fine to send a document without looking at what is inside it.

That asymmetry is the whole point. You cannot control which jurisdiction’s rule the person on the other side follows. You can control what you send.

The five fields that matter in a filing

A PDF stores a block of descriptive data called the Info dictionary. Nine fields in it are defined by the specification. Five of them are worth a look before anything goes to a docket.

Field What it says about the filing
Author The account name on the machine that created it, which is often the drafter’s full legal name
Creator The program the document was written in, such as Microsoft Word
Producer The program that made the PDF
CreationDate When the document was first made, to the minute
ModDate When it was last changed, to the minute

The names are the obvious risk. A declaration meant to read as a witness’s own words that lists an associate as Author is a fact about the document that you did not intend to produce. So is a brief whose Author is a lawyer who is not on the signature block.

The dates are the quiet risk. Between them, CreationDate and ModDate describe the drafting timeline. A document that was supposedly prepared over two weeks and carries a CreationDate forty minutes before it was signed is a discrepancy that costs nothing to find.

The fields nobody shows you

Here is the part that catches careful people out. Those nine fields are the ones the specification defines. Nothing stops a program from writing extra fields of its own, and Microsoft Word does exactly that. A PDF exported from Word commonly carries a Company field and a SourceModified field. Neither is in the standard, so neither is displayed in the document properties panel of Acrobat, Preview, or a browser viewer.

You open the properties box, see a blank Author, and conclude the file is clean. Your firm name is still in the file. We covered the mechanism in detail in what is hidden inside your PDF.

There is a second copy of much of this as well. Modern PDFs carry an XMP packet, an XML block that duplicates the descriptive data. Clearing the Info dictionary and leaving the XMP packet alone leaves the information in the file. Several tools that advertise metadata removal clear only one of the two.

Redaction is a separate problem, and a harder one

It is worth saying plainly, because the two get conflated. Drawing a black rectangle over text in a PDF does not delete the text. The rectangle is a shape drawn on top. The characters underneath are still in the content stream, still selectable, and still returned by any extraction library. Text under a box has been produced in real filings more than once.

Proper redaction removes the underlying content, not just the view of it. Acrobat Pro and the dedicated litigation tools do this. A metadata scrubber does not, and ours does not claim to. Treat them as two separate steps on the checklist, because they fail in different ways.

Before it goes to the docket

Filed documents are public. PACER is a public system and a served document is in the hands of someone whose interests are opposed to your client’s. The check is worth five minutes.

  1. Redact the visible content first, with a tool that removes the text rather than covering it. Verify by selecting the redacted area and trying to copy it.
  2. Apply the Rule 5.2 limits to social security and taxpayer numbers, birth dates, minors’ names and financial account numbers. Check your district’s local rules too, since many add categories.
  3. Look at the metadata, including the non-standard fields your viewer hides.
  4. Strip what should not travel. Keep Title if the docket benefits from it. Remove the rest.
  5. Check the file that will actually be filed. Not the draft, not the one before the last export. Every save can write the fields back.

Step 3 and step 4 are the ones with no tool in most firms’ workflow, which is why they are the ones that get skipped.

A note on scanned filings

Many courts require filings to be text searchable, which means running OCR over anything that came from a scanner. OCR adds a layer of invisible text behind the image so the document can be searched. That invisible text is real text. If you scan a document that has a physical redaction on it, OCR reads what it can see, which is fine. If you scan first, OCR, and then draw a box in the PDF, the OCR text under the box is still there and still searchable. Order matters.

Standard caveats

This is a description of how PDF files work, not legal advice. Ethics opinions differ by jurisdiction, local rules add requirements on top of the federal rules, and your obligations are whatever your bar and your court say they are. What does not differ is the file format. The fields described above are in the document whatever your jurisdiction thinks about them.

Further reading