mythsthe wider recordAI hallucination court wave2025legal

everyone assumed:

only careless solo lawyers cite fake AI cases

what was actually true:

AI fabrication is a property of the tool, not a failing of careless users alone.

how it happened

  1. whoAI drafting toolused for legal research
  2. the gapunchecked filingsno one opened the sources
  3. the resulthundreds of casesfines, disqualification, public naming
the gap that let it happen the result

what everyone assumed

After the first embarrassments, the comforting story was that this was a competence problem at the margins, a thing careful professionals at serious firms would never do. The assumption is that expertise and reputation are protection against the tool’s failure mode.

what happened

Through 2025 the documented cases multiplied into the hundreds, and they reached well-resourced firms. In one matter a court did not merely fine the attorneys but disqualified them and ordered the opinion published, with regulators notified. The fabrications did not respect the seniority of the hand that filed them.

step by step

  1. lawyers file AI-drafted briefsCitations look real and go unchecked before filing.
  2. courts find cases that do not existThe fabrications appear from large, well-resourced firms as well as small ones.the assumption breaks here
  3. sanctions escalateFrom fines to disqualification, published opinions and regulators notified.
  4. late 2025trackers log hundreds of casesThe problem moves from the margins to the mainstream.

how it surfaced

By late 2025 trackers logged hundreds of AI-hallucination filings, including from large and respected firms, with sanctions escalating from fines to disqualification and public naming. The problem had climbed from the margins to the mainstream.

AI fabrication is a property of the tool, not a failing of careless users alone.

Expertise does not immunize anyone who trusts unverified output, because the output looks identical whether the user is junior or senior.

picture it at a company like yours

The company here is made up. The other two paths are not: they happened, and each links to the story.

your company
a normal dayYour best people use AI to draft reports, proposals and client memos faster.
the gap nobody seesNobody checks the facts and sources in AI-drafted work before it goes out, because the authors are trusted.AI output nobody checks
  1. in this story
    the day it mattersA proposal reaches a client quoting a regulation and a case study that do not exist.
    what it costs youThe client finds out first, and your most capable person is now the one who has to explain it.
  2. or it goes like this
    howA customer-facing bot promises something your policy does not offer.
    what leavesmoney you now owewho ends up with itthe customer, backed by a tribunal
    it happenedMoffatt v. Air Canada, 2024
  3. or it goes like this
    howA bot is talked into mocking your company and the chat is shared.
    what leavesyour reputation, in screenshotswho ends up with ita million strangers in a day
    it happenedDPD chatbot, 2024

sources

  1. Scientific American / LawSites, 2025

← all myths

your turn

what you can do about it

  1. this week

    Name the step where AI-drafted work gets checked before it ships, whoever wrote it.

    The people in your company using AI for important work are not the careless ones; they are the capable ones moving fast, and the tool fabricates for them just as readily. This wave is the proof that competence is no shield. Whether verification happens before AI output ships, regardless of who produced it, is the control.

  2. go deeper

  3. if you're not sure where to start

    Run the offboarding checklist on one leaver. It covers every account and token to close when someone leaves.

    open the checklist →

the letter · every two weeks

one access gap in your inbox, with the fix.

Every issue takes one way access creeps in, shows a real incident where it went wrong, and ends with one thing to check that week. Written for the people who run the tools, without a security team behind them.

  • 1one mechanism, explained plainly
  • 2one real incident, taken apart
  • 3one step you can take on Monday

get the next issue

One email every two weeks. Your address stays here and is never passed to vendors. One click to unsubscribe. Privacy.