Begin initial service of the Liberty Pipeline

The company cancelled or abandoned the commitment

AI-confirmed · human review in progress Independent AI checks agreed on primary SEC or regulator evidence. Records are now being reviewed by hand, starting with those our customers order.

Announced: 26 Jul 2019
Due: 31 Mar 2021
Withdrawn: 31 Mar 2021

Deadline
Met
Late
Not met
Today

What the company said

“Liberty is supported by long-term shipper commitments and initial service is targeted for the first quarter of 2021”

SourcePhillips 66 Form 8-K press release, 2019-07-26
FilingSEC filing index
Deadline as statedthe first quarter of 2021
Deadline used31 Mar 2021, the last day of the stated period
TypeProject
Permanent referenceKI-C-2019-PSX-010117
Cite as https://keptindex.org/c/KI-C-2019-PSX-010117/ — this address never changes.

What we found

Evidence strength A What the grades mean

The deadline has not passed yet, so there is nothing to check. We look for evidence starting 30 days after 31 Mar 2021.

Commitment lineage

Every time the company restated this commitment. The original is never changed, and the outcome is always measured against version 1.

Deadline drift +91 days
2 versions

Version 1 · original · stated 26 Jul 2019

Deadline 31 Mar 2021 (“the first quarter of 2021”) · wording aspirational
Version 2 · stated 25 Oct 2019

Deadline 30 Jun 2021 (“first half of 2021”) · wording expected

“Liberty is supported by long-term shipper commitments, and initial service is expected in the first half of 2021.”
Source for this version

Commitment DNA

The structured record of this commitment. Fields the company did not state are left blank rather than guessed.

Event type: Project other
Indicative materiality: Low rules
Action: Begin
Deliverable: Liberty Pipeline
Where: Oklahoma
Language: Aspirational (“aims”, “intends”)
In the company’s control: High
Specificity: High
Conditional: Yes: supported by long-term shipper commitments
Depends on: Customer

How this entry was checked

Commitment recordedThe company's words were matched word for word against the filing.
First checkAn AI model, Claude (Anthropic), read the company's later filings and recorded the outcome with a quoted passage.
Second check: agreesA different AI model, Gemini (Google), re-read the evidence without being told the outcome and reached the same answer.
Third check on a negative finding: agreesBecause a negative finding could harm a company if wrong, a stronger model re-read the evidence fully blind and reached the same answer.

Confirmed. This outcome counts towards the company's Kept Score. How verification works

Are you Phillips 66? You can publish an official statement next to this entry, free. Add a statement.

Is something here wrong or out of date? Companies and readers can ask us to check it. Please bring a link to a public source.