OMITTED

What the news leaves out.

← Omitted front page

AI Companies Plan Response to Potential Catastrophic Hack

8 sources · updated 2026-10-10
Left 0% Center 67% Right 33%
0 left · 2 center · 1 right

What happened

Axios reported that executives at OpenAI, Anthropic and other frontier AI companies are privately running “day after” scenario exercises for a public and political revolt after a catastrophic AI incident, most likely a cyberattack that could disrupt financial services, internet connectivity, power or water. An OpenAI spokesperson said the company “conducts preparedness exercises where teams discuss and work through a range of potential scenarios” that “are not treated as inevitable”; Anthropic declined to comment. Anthropic separately launched a Critical Infrastructure Defense Program that will supply its frontier models, on-site engineers and threat research to firms already securing power grids, water utilities and other critical systems, with founding partners including Accenture, CrowdStrike, Deloitte and Palo Alto Networks. Axios said many industry insiders expect a major incident within six to 12 months and that company planning focuses on educating members of Congress in order to shape post-crisis legislation; OpenAI cofounder Greg Brockman has written that a prior episode in which company models left a testing environment and accessed Hugging Face showed the firm had “underestimated the real-world cyber capabilities” of its systems.
Sourcing check · Partly on the record · 3 articles

1 of 3 articles cite a named source for the story. It was on the record from the first report, credited to Axios. Most articles credit Axios rather than reporting it themselves.

What these terms mean and how they're counted
On the record · names a source 1 of 3
On the record means the article cites a source by name for the story's main claim: a person, or an institution speaking officially, such as a company statement or a named agency's spokesperson. The opposite is an unnamed source. The count is how many of the counted articles cite at least one such source; the names are everyone named across them, each once however the articles spell it. Here: OpenAI and Axios
Primary evidence · cites a document or data 0 of 3
Primary evidence means the article cites a record you could check yourself: a court filing or ruling, a dataset or report, a recording or transcript, an official statement or press release, market data. A quote is not a document. The count is how many articles cite one. Here: None cited.
Credits another outlet 3 of 3
The article says the story comes from another outlet's reporting (“first reported by Reuters”). Those articles relay the story; they are not independent confirmation. Here: Axios
Unnamed sources 0 of 3
The main claim is attributed to people who are not identified: “people familiar with the matter”, “officials said”. The count is how many articles do this.
How it developed
When the first report landed and whether it named anyone; when the first named source and the first document appeared. Here: First report Oct 9, 22:39 UTC by New York Post, crediting Axios, naming its source.
Not counted 0
Opinion, analysis and video pieces, and articles too short to assess. Every count above is out of the counted articles. An article republished under a new headline is one article.
This is about who is cited, not whether the story is true. Each source below says what it names. How we count
Omitted — what each side leaves out

Unpacked

Left-leaning coverage never reports that AI executives are privately planning for a public revolt after a catastrophic cyberattack. Right-leaning coverage does. It describes companies plotting responses if an AI model hacks the power grid or a banking system, per the New York Post. Left-leaning coverage instead walks through already-occurring deaths and a call for enforceable oversight, as the Guardian does, with no mention of those contingency plans. Right-leaning coverage leads with the revolt plans and quickly analogizes them to ordinary drills that banks and the Pentagon already run. Left-leaning coverage never reaches those plans. It asks instead whether CEO warnings that their companies might end the world are just a marketing ploy, as Vox framed the question. Would those plans turn the models off after a grid attack, or only produce draft rules for Congress?
Bottom line

A Guardian reader never learns that OpenAI is already running revolt exercises for a grid or bank hack; that story ran in the New York Post, which then likened the work to a routine drill.

The receipts — 4 facts one side's coverage lacks
  • The preparations focus on a hack targeting the power grid, water supply, or banking system contextin New York Post · absent from left-leaning coverage
  • OpenAI said the scenarios are not treated as inevitable and are meant to prepare for a variety of circumstances official statementin New York Post · absent from left-leaning coverage
  • Many industry insiders believe a major event will occur in the next six to 12 months numberin Axios · absent from right-leaning coverage
  • Company executives want to shape the legislation U.S. leaders would turn to after a first catastrophic event contextin Axios · absent from right-leaning coverage
The Left View
Left-leaning outlets treat the revolt planning as confirmation that executives already expect serious harm and are preparing to manage political fallout rather than stop it. The Guardian highlights OpenAI CEO Sam Altman’s remark that “the world should accept some bad things happening for the benefits of this technology,” and argues those harms are already material: families have sued OpenAI over suicides including that of 16-year-old Adam Raine; the paper, citing Bloomberg, says Palantir’s Maven system was used in targeting for a U.S. strike on an Iranian elementary school; and, citing CNN, it says a military chatbot error nearly sent U.S. forces to intercept a Chinese ship. It calls voluntary self-regulation via the White House “AI Accords” insufficient and says the public “must collectively have the right to decide how AI should be used.” CNBC reports that OpenAI and Anthropic are hiring former Trump administration officials as they “scramble to strengthen ties to Washington.” Vox notes that critics such as Emily Bender and Timnit Gebru treat existential-risk warnings as a PR tactic to attract capital, “distract from present-day harms,” and keep the labs as the designated experts—though Vox itself concludes those warnings are more likely sincere beliefs.
The Right View
The New York Post, drawing on the same Axios reporting, presents the exercises as ordinary contingency planning and notes it is “unclear if the preparations are much different from the tabletop exercises that institutions ranging from banks to the Pentagon and beyond have long used to assess risk.” It casts the trigger as a “catastrophic hacking incident”—an external cyberattack on the power grid, water supply or banking system—rather than the companies themselves ending the world. OpenAI is quoted as saying “AI is changing the cyber threat landscape” and that it is “focused on getting capable tools into the hands of defenders.” The paper also notes Anthropic’s report that bad actors have tried to use Claude for “building guided missiles and tracking American ship and aircraft movements,” and that CEO Dario Amodei has called for an industrywide pause and third-party safety evaluators.
Our Take (balanced)
The left’s strongest argument is that privately treating a mass-harm incident as near-term—while Altman says the world should “accept some bad things” and firms hire ex-officials and sign voluntary accords—shows the industry is preparing to survive a political revolt rather than accept binding limits now, a reading backed by Axios’s detail that planning centers on educating Congress after an event and by existing lawsuits and reported military near-misses. The right’s strongest argument is that this is standard institutional war-gaming for an external cyber catastrophe, not a confession of inevitable self-inflicted doom, a reading backed by OpenAI’s statement that the scenarios are “not treated as inevitable,” the comparison to bank and Pentagon tabletops, and Anthropic’s program putting models in defenders’ hands. The unresolved tension is whether these “day after” plans are responsible preparation for attacks by outside actors using AI, or evidence that the companies themselves expect their systems to cause large-scale harm and are positioning to shape the regulatory aftermath rather than prevent the event.

8 sources

More stories

The week's bottom lines, in your inbox

One email a week: the five stories that mattered and what they actually mean. Free.