Skip to content
Regulayer™Human Control for AI
Book a live demo

The law library · United States · proposed federal rule

Proposed Federal Rule of Evidence 707

Machine-generated evidence · published for public comment 10 June 2025 · comment period 15 August 2025 to 16 February 2026 · withdrawn as published, a working draft under study, not adopted

Read against the United States Courts’ own pages and the Advisory Committee’s agenda book for its 15 October 2026 meeting. This page exists mainly to correct a widely published error.

We built this page because at least one law firm has published that Rule 707 was “approved by the Judicial Conference”. It was not. We checked the United States Courts’ own page on 5 August 2026, when Rule 707 was still listed among the proposed amendments published for public comment, and again on 7 September 2026, when it was not: the page now carries the August 2026 cycle, proposed amendments to Evidence Rules 104 and 902, and neither is about artificial intelligence.

Where it actually stands

“On June 10, 2025, the Judicial Conference Committee on Rules of Practice and Procedure (Standing Committee) approved publication of proposed amendments to the following: Appellate Rule 15; Bankruptcy Rule 2002, Official Forms 101 and 106C; Civil Rules 7.1, 26, 41, 45, and 81; Criminal Rule 17; and Evidence Rule 609 and new Rule 707. The comment period was open from Aug. 15, 2025 to Feb. 16, 2026.”

That sentence is from uscourts.gov, and the whole correction is in one word. The Standing Committee approved publication for comment. That is the second of six stages, not the last.

  • 1Advisory Committee on Evidence Rules proposes.
  • 2Standing Committee approves publication for comment. Done, 10 June 2025.
  • 3Public comment. Closed 16 February 2026.
  • 4Advisory Committee considers comments and decides whether to recommend final approval.
  • 5Standing Committee, then the Judicial Conference, then the Supreme Court.
  • 6Lies before Congress, then takes effect, ordinarily on 1 December.

Where it stands now

The published proposal has been withdrawn. The agenda book for the Advisory Committee on Evidence Rules’ meeting of 15 October 2026, posted by the United States Courts on 23 September 2026, records that after testimony at two hearings and more than 70 written comments, the Committee at its last meeting “voted to withdraw the proposed rule as issued for public comment, and to submit a modified rule proposal for further study.”

The working draft now under study presumes an expert. Where AI output is offered without an expert witness, the proponent would have to show it meets the Rule 702 reliability factors, and “Ordinarily, the proponent must provide an expert to explain how the system of artificial intelligence reliably produced the evidence. But in exceptional circumstances, the court may consider other proof”. It adds a notice requirement. The agenda book says the Committee has not approved all of the draft’s language.

The deepfakes rule is not approved either. The Committee has prepared, “but not finally approved”, a new Rule 901(c) under which the opponent would first have to produce evidence sufficient to support a finding that an item was a deepfake, and the proponent would then have to show that the item is more likely than not authentic. Expert panels on both, AI evidence and deepfakes, were set for the morning of 15 October 2026.

The Committee met in Boston on 15 October 2026 with both panels, and the Rule 707 mini-conference, on its agenda.

So there is no effective date. The rule has not been approved by the Judicial Conference, it is no longer on the Judiciary’s comment page, and the review is continuing.

On 14 August 2026 the Judiciary opened its next comment cycle. The proposed amendments are to Evidence Rule 104(a) and (b), on the standard of proof for preliminary questions and conditional relevance, and Evidence Rule 902(1)(A), on the self-authenticating public records of federally recognised Indian Tribes and Nations, with comments open to 15 February 2027. We read the 209-page preliminary draft in full. The words “artificial intelligence”, “machine-generated”, “deepfake” and “Rule 707” do not appear in it. The published amendment to Rule 104(b) concerns conditional relevance; it does not address deepfakes or machine-generated evidence. The one thing every stage of this record agrees on is the narrow point we verified first: Rule 707 is not in force, and no date for it exists.

Why the error matters

Because people are planning around it. A firm that believes Rule 707 is adopted may be telling clients that machine-generated evidence is already subject to a federal reliability standard. It is not. Meanwhile the actual position is the more interesting one: the United States has no rule on deepfake or machine-generated evidence, while Qatar has had one since January 2026 and New South Wales has regulated AI in expert reports since February 2025.

The gap is not going to be filled by a federal rule on any published calendar: the proposal is back with its drafters for revision and further study. It is being filled in the meantime by individual judges’ standing orders and by the sanctions record.

What to do about it

Or run it across the estate.

While there is no federal rule, the question is still being asked, case by case, by judges who will not wait for one. What they ask for is the same in every jurisdiction: what the machine produced and what a person checked. The same engine either way: a signed record made on your own machine, content-free, and checkable free by anyone, forever.

Or Regulayer™ for enterprise →

See also the attorney certification, the free verifier, and every jurisdiction we track.

Sources

Related

Status read at source, at the Advisory Committee’s agenda book for 15 October 2026. Information, not legal advice.

Common questions

Has Federal Rule of Evidence 707 been approved?

No. Rule 707 was published for comment on 10 June 2025, with the comment period running from 15 August 2025 to 16 February 2026, and it was never approved. On 14 August 2026 the United States Courts' own page moved on to a new comment cycle, proposed amendments to Evidence Rules 104 and 902 with a comment period to 15 February 2027, and Rule 707 is not on it. The 209-page preliminary draft for that cycle does not mention artificial intelligence, machine-generated evidence or deepfakes. Approval for publication was never adoption, and the proposal has now dropped out of the publication cycle altogether.

What would Rule 707 do?

It addresses machine-generated evidence: output produced by a process or system, offered without a human expert witness, and would subject it to the reliability standards that apply to expert testimony. The working draft under study since the published version was withdrawn would ordinarily require an expert, with other proof only in exceptional circumstances. It has been considered alongside a separate draft on deepfakes, Rule 901(c), which is not approved.

When could Rule 707 take effect?

There is no effective date. The Advisory Committee on Evidence Rules has withdrawn the proposal as published for comment and is studying a modified working draft, with an expert panel at its meeting on 15 October 2026. A federal rule must still be approved by the Advisory Committee, the Standing Committee, the Judicial Conference and the Supreme Court, and then lie before Congress. Any page publishing an effective date for Rule 707 is publishing a guess.

Is there any rule on deepfake evidence in the United States?

Not yet at federal level. By contrast, Qatar's QICDRC Practice Direction No. 1 of 2026 has required since 6 January 2026 that visual or audio evidence generated or enhanced using multimedia AI be disclosed with its source, and provides that suspected undisclosed deepfakes may be excluded and investigated.

Information, not legal advice, and not advice on the law of any jurisdiction. Every entry is verified against the issuing body’s own document; where a source is reporting rather than the document, we say so.