“If any of these lawsuits prevail, it would be calamitous for OpenAI and Anthropic, and even more so for Google and Meta, whose Gemini and Llama models use datasets including millions of artists' work, mostly because it's "virtually impossible" for AI models to forget training data, meaning that they'd need to be retrained from scratch, which would cost billions of dollars and dramatically reduce their efficacy at tasks they're already not particularly good at.”
Our reading, the sentence that was checked:
If any of these copyright lawsuits prevail, it would be calamitous for OpenAI and Anthropic and even more so for Google and Meta, mostly because models can't practically forget training data and would need costly retraining from scratch that would cost billions and dramatically reduce their efficacy.
Assessment: Unresolved No deadline stated on 2026-10-29 Evidence as of
Where and when Ed Zitron wrote it
Where's Your Ed At, The Subprime AI Crisis, paragraph 58.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment wyea-subprimeai-9a294c-a2
We recorded 50 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 2024-09-16. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Conditional forecast
- Deadline, as the speaker wrote it
- None stated. This site does not add one.
- Deadline state when assessed
- No deadline stated
- Deadline state on 2026-10-29
- No deadline stated (the release’s reference date)
Conditions
- if any of these lawsuits prevail (antecedent, stated by the speaker)
From the speaker
- If any of these lawsuits prevail
- calamitous for OpenAI and Anthropic
- even more so for Google and Meta
- retrained from scratch
- cost billions of dollars
- dramatically reduce their efficacy
Our interpretation
- The preceding discussion makes a plaintiff victory establishing copyright liability for model training the relevant trigger. Surviving dismissal is insufficient.
- The nearest referent is the litigation involving Stability AI, DeviantArt and Midjourney. Whether the phrase also includes every copyright case mentioned in the post is unclear.
- A settlement concerning acquisition of pirated books does not automatically establish the stated training-liability trigger.
- The practical prediction requires evidence of mandatory complete retraining, its cost, and substantial capability losses. A financial penalty alone does not establish that chain.
Counts as supported if: A qualifying plaintiff victory leads to mandatory retraining from scratch, documented multibillion-dollar retraining costs, substantial capability losses, and the claimed severe consequences for the named developers.
Counts as contradicted if: After a qualifying victory, a final remedy and documented implementation establish that the affected models can continue operating without the predicted complete retraining and capability losses.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Did the forecast come true?
Outcome: Unresolved
The record was checked, and the evidence available by the cutoff does not settle it.
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 deadline stated
- Assessment recorded
- Policy version
- 2026-10-28.7
Rationale
The February 10, 2026 Andersen order resolved a discovery dispute, without establishing copyright liability for model training [e1]. A case tracker reviewed on October 12, 2026 described the case as active and reported further discovery [e2]. The July 20, 2026 Anthropic settlement approval rejected model-deletion requests as outside the settlement's scope and provided for destruction of pirated works [e3]. That settlement does not establish the required merits victory on training liability [e3]. The supplied evidence therefore establishes neither the qualifying victory nor the implemented remedy needed to assess the predicted complete retraining, multibillion-dollar costs, capability losses and business consequences [e1] [e2] [e3].
Limitations
- The claim specifies no deadline and does not exhaustively identify the covered lawsuits.
- The later Andersen status comes from a secondary tracker rather than a complete court docket [e2].
- The settlement order does not establish remedy implementation following a qualifying training-liability judgment [e3].
- The packet provides no matching retraining expenditures, before-and-after capability measurements or comparative business consequences for the named developers.
Evidence (3 items)
Context only: U.S. District Court for the Northern District of California,
No further responses
Why it matters here: This order resolves discovery responses in Andersen. It does not establish copyright liability or order model retraining.
- Publisher
- U.S. District Court for the Northern District of California
- Kind of source
- Regulatory filing
- Describes events of
- Public from
Context only: AI Lawsuit Tracker,
further discovery questions
Why it matters here: The later status summary still describes discovery and pending merits proceedings. This supports withholding a finding that the required plaintiff victory occurred.
- Publisher
- AI Lawsuit Tracker
- Kind of source
- Secondary summary
- Describes events of
- Public from
Context only: U.S. District Court for the Northern District of California,
These requests
Why it matters here: The court rejected requests including deletion of Anthropic's models; the settlement instead includes destruction of pirated works. This challenges treating every plaintiff recovery as necessarily requiring model destruction, but does not establish the narrower training-liability trigger.
- Publisher
- U.S. District Court for the Northern District of California
- Kind of source
- Regulatory filing
- Describes events of
- Public from
What would settle it
A qualifying merits victory establishing copyright liability for model training, followed by a final remedy and sufficiently complete implementation showing whether complete retraining was required, its cost, capability changes and consequences for OpenAI, Anthropic, Google and Meta.
Where it was said (1)
Said once in the sources this register covers.
: Where's Your Ed At, The Subprime AI Crisis
If any of these lawsuits prevail, it would be calamitous for OpenAI and Anthropic, and even more so for Google and Meta, whose Gemini and Llama models use datasets including millions of artists' work, mostly because it's "virtually impossible" for AI models to forget training data, meaning that they'd need to be retrained from scratch, which would cost billions of dollars and dramatically reduce their efficacy at tasks they're already not particularly good at.
- Where
- paragraph 58
- Published
- 2024-09-16
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment wyea-subprimeai-9a294c-a2: Current. Outcome: Unresolved. Evidence as of 2026-10-28. Public date not recorded.
Recorded events
Assessment publishedBy this site
First assessment published after both reviews and the final-text check passed.
HeldBy this site
Both reviews say forecast_outcome/open_horizon/unresolved; the drafted assessment says forecast_outcome/open_horizon/not_adjudicable.
About assessment wyea-subprimeai-9a294c-a1, which was not published.