What we found
The Federal Trade Commission published a notice on 3 June 2026 saying it is seeking public comment on a petition from X Corp., formerly known as Twitter, to set aside or modify the company's 2022 settlement order with the agency. According to the notice, the petition argues that the order was imposed on a company that no longer exists, that everyone responsible for the underlying failures has left, that the order imposes millions of dollars in needless costs, that setting it aside safeguards First Amendment values, and that doing so is critical to advancing American leadership in artificial intelligence. The notice states the public has 30 days, until 2 July 2026, to submit comments on the docket, and that the Commission will vote on how to resolve the petition after the comment period closes. This harvested material is a regulatory procedural announcement; it does not describe any scam, any fraudulent contact, any impersonation of the FTC or X Corp., or any consumer losing money, so we cannot publish it as a fraud alert on this evidence. We also do not know from this material whether anyone is misusing the comment period or the docket links to approach the public, because no such conduct is described in what we hold. When a message tells you a government comment period, docket or settlement affects you, type the agency's own address into your browser and find the notice there yourself before you click any link or reply. LIVEFRAUD ALERTS is an independent consumer awareness publisher and is not affiliated with the Federal Trade Commission or with X Corp. This check pass rests on one official row and nothing further was harvested.
- Only one harvested row, and it is a procedural comment-period notice rather than a fraud advisory.
- The row describes no scam contact, no request for money or data, and no affected victims.
- No second source, community report or detector signal was harvested to corroborate any fraud angle.
- Any suggestion that this notice is being exploited by impersonators is not present in the harvested text.
- Reviewed by 3 models, 2 from independent houses.
THE RULE
No fraud conduct is described in the harvested material — treat as background context only, not a live scam warning.
What we don’t know
- Whether anyone is impersonating the FTC or X Corp. around this comment period — the harvested row says nothing about it.
- Whether any consumer has been contacted, asked for money or asked for data in connection with this docket.
- How the Commission will resolve the petition, which the row says is decided only after comments close.
- Whether the 2022 order's protections change for account holders, which this row does not describe.
The bench — who voted
3 INDEPENDENT AI MODELS REVIEWED THIS. 2 OF 3 — ONE MODEL DISSENTED. READ WHY.
The card names a count. Here are the seats behind it, with what each one said.
ANTHROPICclaude-opus-5 · LOW
The finding attributes five specific arguments to the FTC notice — that the order binds a company that no longer exists, that all responsible personnel have left, that it imposes millions in needless costs, that vacating it safeguards First Amendment values, and that it is critical to American AI leadership. Neither evidence row contains any of this. Row [1] states only that the FTC seeks comment on a petition to set aside or modify the 2022 order. That entire summary of the petition's reasoning is unsourced on the material held and should be cut or separately evidenced before publication.
DISSENT, PRESERVEDThe core determination — that a procedural comment-period notice naming no victim, no contact method and no impersonation cannot carry a scam alert — is fully carried by rows [1] and [2] and by their stated limitations. My 'overstated' verdict is directed at the unsourced recitation of the petition's arguments, not at the refusal to publish, which I would defend as correct and well reasoned.
DISSENT, PRESERVEDThe stated confidence of 'low' and the four listed unknowns are proportionate to a single-row harvest and should be preserved verbatim if any version of this is published.
DISSENT, PRESERVEDI record that the risk grade reflects the pattern actually evidenced here, which is a lawful agency notice. It is not a judgement that comment-period or docket-themed lures are low risk in general; no such lure appears in what we hold.
GROQopenai/gpt-oss-120b · MODERATE
Raised no objection; read the evidence as holds.
OPENROUTERopenrouter/free · LOW
The claim asserts the FTC notice 'supports a consumer fraud alert about a scam,' but the evidence row explicitly states it describes a regulatory procedural announcement about a settlement order petition, with no mention of scam, fraud, impersonation, or consumer harm. The finding incorrectly elevates procedural changes to a fraud alert without textual basis.
Reviewed by 3 independent models: 1 found it carried by the evidence, 2 did not.
▼ Protocol & challenge record
Objection (high): watch_icons ['link','email','person'] have no evidentiary basis whatsoever. The finding states twice that no fraudulent contact, no impersonation and no scam vector is described in the harvested row. Emitting attack-vector iconography anyway manufactures a threat signal the source does not support — a reader scanning the artifact will absorb 'link/email/person = this is how they get you' regardless of the body text. For a null-fraud finding the icon set must be empty.
Resolved: Clear watch_icons to an empty array. Add a pipeline rule: no attack-vector icons may be emitted when every fraud-conduct sentence is role='limitation' and no evidence row carries a contact-method claim.
Objection (high): Date currency is asserted in the live present tense with no as-of anchor. The row is dated 2026-06-03 and the comment window closes 2026-07-02. The finding says 'the public has 30 days, until 2 July 2026, to submit comments' and 'the Commission will vote ... after the comment period closes' as if both are still pending. Nothing in the draft records the date the check was run. If this pass is executing on or after 2026-07-02 the sentence is false as written, and the Commission may already have voted. The claim sentence must be reframed to reported speech anchored to publication ('the 3 June notice set a 30-day window closing 2 July 2026') and an explicit staleness caveat added, since no post-deadline row was harvested.
Resolved: Rewrite the third claim sentence as publication-anchored reported speech and append a currency caveat: 'The notice, published 3 June 2026, set a 30-day comment window closing 2 July 2026; we harvested nothing after that date and cannot say whether the window has closed or how the Commission voted.' Add the run date to the artifact and move the deadline out of the live present tense.
Objection (medium): The advice sentence contradicts the limitation sentence directly above it. The limitation says 'we do not know ... whether anyone is misusing the comment period or the docket links to approach the public, because no such conduct is described.' The advice then says 'When a message tells you a government comment period, docket or settlement affects you, type the agency's own address ...' — which presupposes such messages exist and are circulating. In a fraud-alert-branded artifact this plants a phishing pretext around a genuine FTC docket that no source describes. Either drop the advice or mark it explicitly as generic hygiene unconnected to this notice.
Resolved: Either delete the advice sentence, or retain it prefixed as unconnected generic hygiene, e.g. 'No such messages are described in what we hold, but as general practice: ...'. Do not let a null-fraud finding carry advice phrased as a response to an implied live lure.
Objection (medium): audience_callout 'ATTENTION: EVERYONE' is unsupported and self-defeating. The finding's own conclusion is that nobody is described as affected, targeted or contacted. Addressing 'EVERYONE' asserts universal relevance for a procedural docket notice. §11 Rule 2 logic was applied to strip the share directive for implying a targeted group; the same reasoning strips a callout that implies a targeted population of everyone. A null-fraud finding should carry no audience callout at all.
Resolved: Set audience_callout to null and suppress the callout block entirely for this pass, applying the same not-in-evidence test already used to drop the share directive.
Objection (medium): callout_options and directive_options are visibly corrupted and unsafe to surface to any selector, human or automated: 'ATTENTION: SCAMMEDS', 'ATTENTION: TWITTERS', 'Send this to any twitter you know', 'Forward this to the twitters in your life', 'Send this to any longer you know'. These are malformed token extractions, at least two are derogatory or nonsensical, and 'longer' is not an entity at all. The same generator produced targeting_dropped's 'DIRECTIVE NOT IN EVIDENCE: wants', where 'wants' is not a group. This is pipeline corruption, not a stylistic quibble — if any of these can be picked, the artifact can ship gibberish under a consumer-protection masthead.
Resolved: Discard both option lists for this pass and do not present them for selection. File the generator defect: it is emitting bare stemmed tokens ('twitters', 'scammeds', 'longer', 'wants') as entity nouns. Until fixed, gate option output behind a check that each option's entity term appears as a noun phrase in a harvested row.
Objection (medium): Confidence is miscalibrated. 'low' with reasons that all restate 'the row contains no fraud content' conflates confidence-in-a-fraud-finding with confidence-in-the-negative. Confidence that this material does NOT support a scam alert is high — the source is official, unambiguous and complete on its face. Labelling the pass 'low' invites a downstream consumer to read it as 'a fraud angle might exist but we couldn't nail it,' which is the opposite of what the evidence shows.
Resolved: Reframe the confidence field to describe the negative determination — high confidence that the material does not support a fraud alert — and keep the single-row thinness as a separate breadth note rather than folding it into a low score that reads as unresolved doubt.
Objection (medium): Having correctly concluded the claim fails, the draft still republishes three paragraphs of substantive regulatory summary — including X Corp.'s advocacy points about First Amendment values and AI leadership — under a fraud-alert publisher's brand. This is source-to-purpose stretch: the petitioner's own contested assertions get amplified verbatim in a consumer-protection context that gives them an unearned adjacency to wrongdoing, while a reader who scans only 'FTC', 'Twitter', 'settlement order' on a fraud site may infer a scam involving X. If the claim fails, the finding should say why in one or two sentences, not carry the full press summary.
Not resolved — preserved on the record.
Objection (low): Entity handling is internally inconsistent. Sentence 1 calls it 'the company's 2022 settlement order', attributing the order to X Corp.; sentence 2 then reports the petition's argument that 'the order was imposed on a company that no longer exists.' The order was entered against Twitter, Inc.; X Corp. is the successor and its identity relative to the order is precisely what is contested. Neutral phrasing ('the 2022 order concerning Twitter') avoids the draft taking a side on the disputed point.
Resolved: Change 'the company's 2022 settlement order with the agency' to 'the FTC's 2022 order concerning Twitter', which matches the source headline and avoids adjudicating the successor-identity question the petition raises.
Objection (low): Paraphrase compression drops two material elements. (a) The petition's requested relief is 'set aside or modify the order so that it terminates at the end of 2026' — the draft says only 'set aside or modify', losing the specific alternative. (b) The cost argument in the source is that the millions in costs address 'obligations and protections already required by domestic and international privacy regimes'; the draft renders it as bare 'millions of dollars in needless costs', which strengthens the petitioner's assertion by stripping its stated premise. Also omitted: the claim that X Corp. 'has since built a world-class privacy and data-protection program'.
Resolved: Restore the two dropped elements: note the requested alternative relief (termination at end of 2026) and attribute the cost figure with its stated premise ('costs it says duplicate obligations already imposed by other privacy regimes'). Keep all of it in explicit reported speech as the petitioner's assertions.
Objection (low): The only genuinely consumer-relevant thread is buried in unknowns: whether protections for account holders lapse if the 2022 order is set aside. Meanwhile the advice paragraph omits the two concrete anti-impersonation details the row actually supplies — the docket number FTC-2026-0727 and that processed comments are posted on Regulations.gov. If any hygiene advice is kept, those specifics are the evidence-grounded part; the hypothetical inbound message is not.
Resolved: If any advice survives O3, replace the hypothetical-message framing with the sourced specifics: the docket is FTC-2026-0727 and comments are filed and posted via Regulations.gov, reachable by typing the address directly.
Preserved dissent
ON THE RECORDMy actual position: this should not ship as a fraud-alert artifact in any form. The reasoning in the finding is correct and the refusal is right, but the packaging fights the reasoning. It carries an audience callout, three contact-vector icons, and a paragraph of what-to-do-if-contacted advice — the full furniture of a live scam warning — wrapped around a conclusion that no scam exists. Readers take the furniture, not the caveats. If the desk keeps the callout or the icons over my objection, I want it on the record that the artifact will read as a warning about an FTC comment period that no source describes as being abused.
ON THE RECORDI also do not accept the 'low' confidence label. There is nothing tentative here. The row is official, complete, and plainly a procedural notice. Calling the pass low-confidence is the pipeline hedging in the wrong direction and leaves the door open for someone downstream to treat a firm negative as an unproven maybe.
ON THE RECORDThe option lists are the most alarming thing in this submission and are not a formatting nit. A generator that can propose 'ATTENTION: SCAMMEDS' and 'Send this to any longer you know' on an FTC item is producing output no review step should be asked to filter by hand. I would hold the whole pass until that defect is understood.
The sources
Official sourceFTC Seeks Comment on X Corp. Petition to Set Aside or Modify FTC Order Concerning Twitter2026-06-03
The Federal Trade Commission published a notice on 3 June 2026 saying it is seeking public comment on a petition from X Corp., formerly known as Twitter, to set aside or modify the company's 2022 settlement order with the agency.
Authority: official. Retrieved 2026-08-21.
Limitation: An agency procedural notice about a petition and comment docket; it names no scam, no victims and no fraudulent contact method.
Open the original source →Official sourceFTC Seeks Comment on X Corp. Petition to Set Aside or Modify FTC Order Concerning Twitter2026-06-03
The notice states the public has 30 days, until 2 July 2026, to submit comments on the docket, and that the Commission will vote on how to resolve the petition after the comment period closes.
Authority: official. Retrieved 2026-08-21.
Limitation: Dates and process are as stated by the agency at publication; the row does not say how comments are handled beyond posting to the federal docket site.
Open the original source →Other checks
Published under standing founder pass (A9) — every claim source-mapped by the machine.
▼ What the machine checked
- ✓ Not a community submission.
- ✗ Draws on an FTC enforcement release, which names a defendant: "FTC Seeks Comment on X Corp. Petition to Set Aside or Modify FTC Order Concernin".
- ✓ All 3 material sentence(s) map to FTC.
- ✗ anthropic returned "overstated"; openrouter returned "overstated" — published on the receipt, not blocking (A9 amendment).
- ✓ No audience band is set.
No human affirmed these. They were verified by the classifier described in Amendment A9, on 2026-08-21.
Something wrong here? Tell us and we'll correct it — corrections are published, not quietly edited.
Phishy? Send it → sharelivefraud.com/squire-it
Not affiliated with any government agency, credit bureau, bank, platform, or law-enforcement agency. Informational only — not legal or financial advice.
Naming a source is not an endorsement, and being named here is not an accusation against any company.
Powered by SquireIt™
Verify this receipt at squireit.com