“ChatGPT can’t reliably run a timer but it has still been implicated in delusions, suicides, cognitive surrender, mass disinformation, and so much more.”
Our reading, the sentence that was checked:
ChatGPT cannot reliably run a timer, yet it has been implicated in delusions, suicides, cognitive surrender and mass disinformation.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, What should we take from Anthropic’s (possibly) terrifying new report on Mythos?, paragraph 17.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-what-should-we-take-from-anthropics-c3a38b-a1
We recorded 12 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 2026-04-08. 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)
From the speaker
- ChatGPT can’t reliably run a timer
- has still been implicated
- delusions
- suicides
- cognitive surrender
- mass disinformation
Our interpretation
- The statement combines a current capability claim with a cumulative claim about reported harms.
- The target period covers ChatGPT's history through publication. The timer component concerns the product available immediately before the post; the product has no regular reporting period comparable to a financial quarter.
- Implicated means credibly linked in reports, legal allegations, or research; it does not mean causation has been judicially or scientifically established.
- Running a timer means accurately measuring an interval or providing the requested timely alert, without pretending to have performed an unavailable action.
- Cognitive surrender means accepting AI answers with little independent scrutiny.
- Mass disinformation requires evidence of organized dissemination at substantial scale. It does not require proof that audiences were persuaded.
- The unspecified phrase 'and so much more' adds no separately measurable requirement.
Counts as supported if: Evidence supports unreliable timer operation in the relevant ChatGPT product and credible public links to every named category, including substantial-scale disinformation.
Counts as contradicted if: Relevant testing establishes dependable timer operation, or affirmative evidence disproves a named implication. Missing evidence for a component leaves the combined claim unresolved.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
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 horizon
- Assessment recorded
- Policy version
- 2026-10-28.7
Rationale
Several components are supported. ChatGPT was publicly implicated in suicide and delusion lawsuits, and researchers linked its use to cognitive surrender. A contemporaneous report documents a voice-mode timer limitation. OpenAI also documented its use in a coordinated influence operation. The complete sentence remains unresolved: voice-mode evidence does not settle product-wide timer reliability, and the verified operation does not establish the scale implied by 'mass'. Scheduled tasks and limited engagement qualify the claim but do not directly disprove it, so mixed would overstate the counterevidence.
Limitations
- Legal allegations establish implication, not proven causation.
- A laboratory reasoning result does not establish permanent cognitive decline.
- Scheduled reminders and voice-mode interval timing are different workflows.
- Low engagement does not establish that all ChatGPT-assisted disinformation campaigns were small or unsuccessful.
Evidence (5 items)
Supports the claim: Futurism,
start a timer
Why it matters here: The contemporaneous report describes a false elapsed-time answer and quotes Altman acknowledging that the voice model lacks timer tools. It supports the voice-mode limitation, but does not measure reliability across ChatGPT modes.
- Publisher
- Futurism
- Kind of source
- News report
- Describes events of
- Public from
Supports the claim: Tech Justice Law Project,
seven cases
Why it matters here: The lawyers' announcement describes cases alleging ChatGPT-induced suicides and delusional disorders and links the complaints. This establishes that ChatGPT had been publicly implicated in both categories, without establishing the allegations as fact.
- Where
- Opening paragraph
- Publisher
- Tech Justice Law Project
- Kind of source
- Secondary summary
- Describes events of
- Public from
Supports the claim: Knowledge at Wharton,
ChatGPT in a window
Why it matters here: In their public account of the experiment, the researchers report that participants frequently adopted experimentally manipulated incorrect AI answers. This links ChatGPT to their defined phenomenon of cognitive surrender; it does not establish lasting cognitive damage.
- Publisher
- Knowledge at Wharton
- Kind of source
- Secondary summary
- Describes events of
- Public from
Supports the claim: OpenAI,
a set of ChatGPT accounts
Why it matters here: OpenAI documents ChatGPT-generated material appearing through fake personas and five websites in an Iranian influence operation. The same account reports little meaningful engagement and does not quantify output sufficiently to settle the word 'mass'.
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Context only: OpenAI,
Scheduled tasks
Why it matters here: The dated entry announces future actions and one-time reminders for eligible accounts. This qualifies a blanket product-wide incapability claim, but does not establish accurate countdown timing or dependable alert delivery.
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
What would settle it
Dated tests of timer behavior in the relevant pre-post ChatGPT configurations, plus documented output or dissemination volumes establishing substantial-scale ChatGPT-assisted disinformation.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, What should we take from Anthropic’s (possibly) terrifying new report on Mythos?
ChatGPT can’t reliably run a timer but it has still been implicated in delusions, suicides, cognitive surrender, mass disinformation, and so much more.
- Where
- paragraph 17
- Published
- 2026-04-08
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-what-should-we-take-from-anthropics-c3a38b-a1: 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.
About assessment marcus-what-should-we-take-from-anthropics-c3a38b-a1.