Pundit ClaimsAI claims by Gary Marcus and Ed Zitron, with the evidence

“Two weeks later, it has already come out that the database that indexes the whole thing is in plain text (except on systems that are explicitly set to encrypt everything). That’s bananas. Keystone Kops level security.”

The misguided backlash against California’s SB-1047, paragraph 37. The full statement is at the source.

Our reading, the sentence that was checked:

It has come out that the database indexing Microsoft Recall's screenshots is stored in plain text, except on systems explicitly set to encrypt everything, which Marcus calls bananas and Keystone Kops level security.

Assessment: Supported Evidence as of

Where and when Gary Marcus wrote it

Marcus on AI, The misguided backlash against California’s SB-1047, paragraph 37.

Said
Source observed
live page, full text, public
Evidence as of
the cutoff for this assessment
Assessment recorded
assessment marcus-the-misguided-backlash-against-californias-c3bf59-a3

We recorded 9 claims from this post. See every claim from this post.

The wording above is the version this site observed on 2026-10-28. The source may have been edited after 2024-06-07. This page does not claim the wording was captured then.

What the claim commits to

Kind
Observation
Deadline, as the speaker wrote it
None stated. This site does not add one.
Deadline state when assessed
No horizon
Deadline state on 2026-10-29
No horizon (the release’s reference date)

Conditions

  • except on systems that are explicitly set to encrypt everything (qualifier, stated by the speaker)

From the speaker

  • Two weeks later, it has already come out
  • the database that indexes the whole thing is in plain text
  • except on systems that are explicitly set to encrypt everything

Our interpretation

  • Assess the early Recall implementation examined around the post date, not later redesigned releases.
  • Plain text means readable, unencrypted database contents, not that the SQLite file is literally a plain-text document.
  • Read the exception as referring to whole-disk encryption, which is distinct from encryption applied to the database itself.
  • Use May 24 through June 7, 2024, reflecting the post's recent two-week context. Its reference to March 24 conflicts with that context.
  • The descriptions bananas and Keystone Kops level security are opinions; the assessable proposition is the database's storage protection.

Counts as supported if: Direct technical examination of the relevant early Recall implementation establishes that its screenshot-derived text index was readable in an SQLite database without separate database encryption, while system-level disk encryption could protect the stored files.

Counts as contradicted if: Direct examination establishes that the relevant early implementation independently encrypted its index database, or that the plaintext reports concerned a different product or materially different implementation.

Retrospective: these criteria were written after the outcome was public.

Current assessment

Was it true for the period described?

Outcome: Supported

The dated evidence matches what the claim committed to, under the criteria frozen before grading.

Scored on a plain reading. The speaker named no measure for this claim, so it was judged on the reading an informed reader would take, stated below under our interpretation. Both reviews had to find that reading fair.

Evidence as of
Deadline state when assessed
No horizon
Assessment recorded
Policy version
2026-10-28.7

Rationale

TotalRecall's June 6, 2024 extraction example, preserved in documentation updated June 8, demonstrates readable screenshot-derived text in an SQLite database without separate database encryption [e1]. Independent testing published June 4 verified access to raw Recall data in Windows 11 24H2 build 26100.712 [e2]. Whole-disk encryption could protect the stored files, but Ars Technica reported that it was generally enabled by default when signed into a Microsoft account [e2]. This supports disk-level protection without establishing that it required a separate manual choice [e2].

Limitations

  • The tests concern an early preview implementation; they do not establish the protections in later Recall releases [e1, e2].
  • The evidence establishes readable SQLite contents, not that the database was literally a plain-text document [e1, e2].
  • The evidence does not establish that whole-disk encryption required explicit manual configuration [e2].
  • The technical evidence does not settle the post's conflicting chronology or its subjective security characterizations.

Evidence (2 items)

  • Supports the claim: Alexander Hagenah,

    unencrypted SQLite database

    Why it matters here: The preserved June documentation describes the unencrypted database and demonstrates extracting and searching its screenshot-derived text. Its June 6 example falls within the target period; June 8 is the documentation's stated FAQ update date.

    Publisher
    Alexander Hagenah
    Kind of source
    Demonstration
    Describes events of
    Public from
  • Context only: Ars Technica,

    Other Windows 11 24H2 versions

    Why it matters here: The author tested build 26100.712 and independently verified access to raw Recall data. The article also explains that whole-disk encryption generally applied when signed into a Microsoft account or using BitLocker, qualifying any suggestion that it always required a separate manual choice.

    Publisher
    Ars Technica
    Kind of source
    Independent measurement
    Describes events of
    Public from

Where it was said (1)

Said once in the sources this register covers.

  • : Marcus on AI, The misguided backlash against California’s SB-1047

    Two weeks later, it has already come out that the database that indexes the whole thing is in plain text (except on systems that are explicitly set to encrypt everything). That’s bananas. Keystone Kops level security.

    Published
    2024-06-07
    Source observed
    , live page
    Access
    full text, public

History

Assessment versions

Recorded events

  1. Assessment publishedBy this site

    First assessment published after both reviews and the final-text check passed.

    About assessment marcus-the-misguided-backlash-against-californias-c3bf59-a3.

  2. HeldBy this site

    The final-text check failed: rationale_supported_by_evidence; overclaims: These findings support the observation about the early Recall index [e1, e2]..

    About assessment marcus-the-misguided-backlash-against-californias-c3bf59-a2, which was not published.