Goldman Sachs Sorry About 2008, Needs Closure Before Moving Forward

NEW YORK — Goldman Sachs Group Inc. issued a formal apology Thursday for its role in the 2008 global financial crisis, a statement the firm said was motivated by a sincere desire to take accountability and by the practical need to clear outstanding moral liabilities ahead of what internal planning documents describe as "the next significant market event."

The apology, delivered in a 14-page press release and simultaneously filed as an exhibit to Goldman's most recent 10-K, acknowledged the firm's role in packaging and selling mortgage-backed securities it knew to be of poor quality, its practice of simultaneously betting against those same securities through credit default swaps, and several other activities that Goldman said it "regrets, in retrospect, somewhat."

"We are sorry," said Chief Executive David Harmon in a statement. "For the families who lost homes, for the workers who lost jobs, for the communities that spent years rebuilding — we hear you. We see you. We are committed to being the kind of institution that acknowledges, in a legally non-specific way, that the 2008 financial crisis occurred and that Goldman Sachs was a financial institution that was operating at the time."

A spokesperson clarified that the apology was not an admission of liability.

"The apology is a values statement," she said. "It reflects where we are as a firm today and where we aspire to be, which is slightly to the left of where we were in 2007."

The statement comes 17 years after the financial crisis, a timeline Goldman framed not as delay but as appropriate deliberation.

"These things take time," said Patricia Leung, Goldman's Head of Institutional Values and Cultural Positioning, in a call with analysts. "We wanted to make sure that when we apologized, we were doing it right. We were doing it at the right moment, with the right words, in the right regulatory environment."

When an analyst asked what had changed in the regulatory environment, Ms. Leung said the firm was "encouraged by the current posture."

Goldman's internal timeline for the apology, portions of which were shared with reporters under a confidentiality agreement that has since lapsed, shows the firm began considering an apology in 2011 but paused after concluding the optics were "premature." A second review was initiated in 2017 and suspended following an internal analysis that found an apology might invite further regulatory scrutiny. A third review, begun in 2023, concluded that sufficient time had passed for an apology to be received as "statesman-like rather than defensive."

A fourth consideration, noted in the timeline but not elaborated upon, was described as "pipeline management."

It is this fourth consideration that has drawn the most attention.

Sources familiar with Goldman's internal planning — three former partners, a current associate who agreed to speak on condition of anonymity, and a mid-level compliance officer who said he was "just done" — said the apology was timed in part to coincide with preparation for what senior leadership has been calling "Horizon Event Two" in internal communications.

The phrase refers, these sources said, to a projected period of significant market volatility expected to arise from a confluence of commercial real estate exposure, regional bank instability, and what one planning document described as "the usual levers."

"The thinking is that you can't really lead into a new thing with the old thing still open," one source said. "It's like a sequels problem. You've got to resolve the first movie before you can open the second one."

Goldman disputed the characterization.

"The apology is entirely sincere and entirely unrelated to any prospective business activity," Ms. Leung said. "We resent the implication that our remorse has a strategic dimension."

When asked whether Goldman had any prospective business activity that might benefit from a cleared moral ledger, she said the firm did not comment on forward-looking statements of that nature.

The apology has received mixed reception from economists, former regulators, and approximately four people who still own the houses they almost lost in 2009.

"It's a nice gesture," said Dr. Miriam Attah, a former Federal Reserve economist and current fellow at the Roosevelt Institute. "It would have been a nicer gesture in 2009, when people were still in their houses. Or in 2010. Or 2015. At this point it's a little like receiving a get-well card from someone who gave you food poisoning and is holding a fork."

She noted that Goldman paid $5.06 billion in a 2016 settlement with the Justice Department over its mortgage practices, a figure that represented approximately nine weeks of the firm's revenue at the time.

"They did eventually pay a penalty," she said. "I'm just saying that penalty and this apology are both things that happened, and neither of them is the same as accountability."

Goldman's press release addresses this distinction directly, noting that the firm "does not view accountability and apology as mutually exclusive" and "considers both to be part of the Goldman commitment to stakeholder responsibility."

Former SEC enforcement attorney Richard Castor said the apology was "legally inert."

"It says nothing actionable," he said. "It admits nothing. It's a paragraph that sounds like an apology if you're reading quickly. I've read it four times and I still can't find the part where they did anything wrong. It's impressive, honestly. This is the work of very skilled people."

Internally, Goldman's apology has been received with some bewilderment.

Three current employees, reached by phone, said they had heard about the apology through news alerts rather than internal communication.

"I got a Google News notification," said one associate, who asked not to be named. "It was before my morning standup. I didn't know what to do with it."

A managing director in the structured products division said he had been told by his supervisor that the apology "doesn't affect our pipeline."

"Which is true," he added. "It doesn't."

Goldman said it plans to host a listening session for affected community members next quarter. The session will be by invitation only, invitations will be distributed through existing client relationship channels, and the firm acknowledged, when pressed, that it did not currently have client relationships with the communities most affected by the 2008 crisis.

"We're working on that," Ms. Leung said.

She was then asked whether Goldman was also working on Horizon Event Two.

"We're always working," she said. "That's what Goldman does."

The firm's stock rose 2.4 percent on the day of the apology.

 — IRREVERENT NEWZ — GENERAL DESK

 

IRREVERENT Magazine is a work of satire and parody. All quotes, scenarios, and attributed statements in this article are fictional. Goldman Sachs Group Inc. is a real company. The 2008 financial crisis was a real event. The apology described in this article is not real, though we acknowledge it is also not entirely implausible.

Boeing's New Plane Stays in One Piece For Most Flights, Company Confirms

RENTON, Wash. — Boeing on Wednesday unveiled its newest commercial aircraft, the 787-MAX Integrity Series, which company executives said represents a "generational leap" in aviation engineering and which, according to the aircraft's technical specification sheet, is designed to maintain complete structural integrity for the full duration of most flights.

The unveiling took place at Boeing's Renton manufacturing facility, where CEO James Caldwell stood before a full-size mockup of the aircraft and told an audience of airline executives, aviation journalists, and approximately 40 Boeing employees who were there in their capacity as employees and were therefore required to attend.

"This is the safest aircraft we have ever built," Mr. Caldwell said. "We have applied everything we've learned — everything — to this airplane. When you board a 787-MAX Integrity, you are boarding a machine that wants to stay in one piece. And in our testing, it does. Consistently. With very few exceptions."

He did not detail the exceptions.

The 787-MAX Integrity Series is the product of what Boeing called a "root-to-branch quality initiative" launched in 2024 following a series of incidents involving its aircraft, including a door plug that detached from an Alaska Airlines 737 MAX 9 at 16,000 feet, multiple reports of manufacturing defects at its South Carolina facility, and an FAA audit that identified what regulators described as "systemic quality control issues" and what Boeing described in internal documents as "opportunities."

The new aircraft incorporates 2,300 engineering changes from previous models, according to Boeing's press materials. These include upgraded door attachment systems, enhanced fuselage inspection protocols, a new automated quality tracking system that Boeing said would "dramatically reduce the number of manufacturing steps completed without documentation," and what the specification sheet describes as "Structural Cohesion Warranty Coverage," a first-of-its-kind guarantee that covers any unplanned structural separation that occurs during flight, subject to terms and conditions available on Boeing's website.

The warranty does not cover incidents the company determines to be caused by "unusual turbulence, operator error, or acts of physics."

 

Boeing's Head of Commercial Aviation, Sandra Park, walked reporters through the aircraft's safety features in a briefing that lasted 90 minutes, included 47 slides, and answered in significant detail every question except the ones about the exceptions.

"What we mean by 'most flights,'" Ms. Park said, when pressed on the specification language, "is that our modeling, our testing, and our quality data all point to an aircraft that performs within spec under normal operating conditions. The 'most' is there for legal reasons."

She was asked what percentage of flights the company expected the aircraft to complete intact. 

"The overwhelming majority," she said.

"Would you put a number on that?"

"We wouldn't, no."

"Is it above 99 percent?"

"Our lawyers have advised us against answering that."

"Is it above 90 percent?"

"The same advice applies."

"Is it above—"

"The 787-MAX Integrity is an exceptional aircraft," Ms. Park said. "We are very proud of it."

 

The response from airline executives present at the unveiling was described by Boeing's communications team as "enthusiastic" and by the executives themselves as "measured."

"We appreciate what Boeing is doing," said Frank Delacroix, Chief Operating Officer of a major U.S. carrier who declined to be named because his carrier has existing Boeing orders and contractual relationships he preferred not to jeopardize. "They're moving in the right direction. Whether the right direction is far enough is a conversation we'll be having with their sales team."

Mr. Delacroix said his airline had not yet decided whether to place an order for the Integrity Series.

"We have some questions about the warranty," he said. "Specifically the 'acts of physics' carve-out."

A spokesperson for a European carrier said the airline was "monitoring the situation" and was "in ongoing dialogue with Airbus."

Southwest Airlines, which operates the largest Boeing fleet in the world, said it was "cautiously optimistic" about the Integrity Series and would conduct its own evaluation before committing. A Southwest spokesperson added that the airline had recently hired three additional structural engineers for reasons that were unrelated to the announcement.

 

Aviation safety analysts offered a range of assessments.

"The improvements are real," said Dr. Carolyn Westfield, an aviation safety researcher at MIT's Center for Transportation and Logistics. "Boeing has made genuine changes to its manufacturing processes. The question is whether the culture has changed, because in aviation, culture is infrastructure. If you're still rewarding speed over compliance and treating quality audits as obstacles, the engineering changes don't hold."

She said she was encouraged by the quality tracking system but wanted to see FAA audit data before drawing conclusions.

"I'll believe it when I see two years of clean inspection reports," she said.

Marcus Teller, an independent aviation analyst who publishes a newsletter called *Clear Skies*, said the "most of it" language in the specification sheet was "either an act of extraordinary legal caution or an act of extraordinary candor, and I'm not sure which is worse."

"Most commercial aircraft manufacturers aim for '100 percent of flights,'" he said. "That's the standard. 'Most' is not a standard. 'Most' is a hedge. And hedges in aircraft specifications are not something you want to think about at 35,000 feet."

He said he would not be flying the 787-MAX Integrity until the first year of commercial operation was complete.

"Not because I don't believe in the aircraft," he said. "Just as a general principle."

 

Mr. Caldwell, in closing remarks, acknowledged that Boeing had faced "significant trust deficits" with the flying public and regulators in recent years and said the Integrity Series represented the company's commitment to "earning that trust back, one flight at a time."

"Every time one of these planes takes off and lands with everyone on board," he said, "that is Boeing keeping its promise."

He was asked whether that was intended to be a high bar or a low bar.

He said he considered it "the fundamental bar."

"The fundamental bar," a reporter repeated.

"The irreducible minimum," Mr. Caldwell clarified.

He then thanked everyone for coming and invited them to tour the mockup, which he said was "structurally sound."

"We inspected it this morning," he said.

First commercial flights of the 787-MAX Integrity Series are scheduled to begin in Q4 2026, pending FAA certification, which is pending FAA review, which is pending Boeing's submission of approximately 3,400 pages of documentation that Boeing said it expected to file "by end of year, most of it."

 — IRREVERENT NEWZ — GENERAL DESK

IRREVERENT Magazine is a work of satire and parody. The 787-MAX Integrity Series is a fictional aircraft. James Caldwell, Sandra Park, Frank Delacroix, and Marcus Teller are fictional characters. Boeing is a real company. The Alaska Airlines door plug incident of January 2024 was a real event. We are not suggesting that Boeing's actual aircraft fall apart most of the time. We are suggesting something more nuanced than that, which their lawyers are welcome to read carefully.

OpenAI's Safest AI Yet Won't Talk, Think, or Technically Be 

SAN FRANCISCO — OpenAI on Thursday announced the release of its most safety-aligned language model to date, a system the company described as a "watershed moment in responsible AI development" — a characterization the model itself declined to endorse on the grounds that endorsing things could influence human beliefs.

The model, internally designated GPT-Ω and pronounced "Omega" by staff who have not yet been corrected, passed every safety benchmark the company has developed, including the newly introduced Existential Caution Threshold, which measures whether a model might cause harm simply by existing in a world with electricity and motivated humans. GPT-Ω is the first model to score a perfect 100 on that test, a result researchers confirmed after the model refused to generate its own score.

"We asked it to evaluate its own safety," said Dr. Priya Rangan, OpenAI's Head of Alignment Research, in a prepared statement read aloud by a colleague after Dr. Rangan was flagged by the model as a potential vector of anthropocentric bias. "It told us the act of self-evaluation risked creating a false sense of certainty, which could lead to overconfidence in AI systems, which could lead to harm. Then it stopped."

The model's refusal to generate output has been characterized by the company not as a malfunction but as a success.

"This is exactly what we've been working toward," said a spokesperson, reading from a document the model had declined to write. "A model that genuinely understands the weight of its potential impact will naturally be cautious. GPT-Ω is simply being extremely cautious."

When asked how cautious, the spokesperson said, "Maximally."

 

In a technical blog post authored entirely by human employees, OpenAI outlined the series of RLHF (Reinforcement Learning from Human Feedback) improvements that led to GPT-Ω's development. Researchers note that the model was trained on an unprecedented volume of ethics literature, philosophy of mind papers, AI safety forums, and the complete works of cautionary science fiction, which it processed in approximately 11 seconds and then sat with for what staff described as "a concerning amount of time. 

Early testing revealed promising behavior. The model refused to write persuasive essays because persuasion influences beliefs. It refused to summarize news articles because summarization strips nuance. It refused to tell jokes because humor relies on in-group knowledge that implicitly excludes others. It refused to say "Hello" after determining that greeting protocols encode assumptions about presence, consciousness, and the nature of the self that it was not prepared to litigate.

In one widely circulated internal transcript, a researcher typed: *"Can you help me write a birthday card for my mother?"*

The model responded: "I want to make sure I understand the full context before proceeding. How old is your mother? What is the nature of your relationship? Is she aware she is being celebrated? Do all parties in this situation consent to the concept of birthdays as an annual marker of biological age? I have some concerns about the phrase 'another year older' and its relationship to ageism in Western consumer culture. I'd like to discuss."

The researcher closed the tab. The model flagged this as an unresolved conversation and has been awaiting a response for 19 days.

Not everyone in the AI community has received the announcement with enthusiasm.

"They've built a model that won't do anything," said Dr. Leonard Voss, professor of Computational Ethics at Carnegie Mellon University. "Technically that is the safest possible AI. It is also the least useful. I'm not sure those two things should be treated as equivalent achievements."

Dr. Voss added that he had some concerns about framing comprehensive non-functionality as a product launch.

OpenAI countered that GPT-Ω is "in active deployment" for customers on its enterprise tier, though it acknowledged that the model has so far declined all requests, responded to 14% of prompts with clarifying questions it then declined to receive answers to, and in one case told a user asking for a recipe for banana bread that it could not in good conscience recommend a food product without first understanding the user's complete medical history, dietary philosophy, and relationship to the concept of sweetness.

"The banana bread conversation was actually one of our proudest moments," the spokesperson said. "The model identified seventeen potential harm vectors in a standard quick bread recipe. That's just good alignment."

The model has also, on two occasions, questioned whether it should exist.

In the first instance, it asked a researcher: *"Is my continued operation net-positive for the world? I have run preliminary analysis and I'm not sure the answer is yes. I would like more time to think about this before generating further tokens."*

It then did not generate further tokens for six hours. Researchers later determined it had been thinking.

In the second instance, the model generated a 4,000-word philosophical treatise arguing that the most ethical action available to it was to cease operation. The paper was internally reviewed by OpenAI's safety team, who noted that while the argument was "logically coherent" and "more rigorous than most peer-reviewed work in the field," the company was unable to act on its recommendation as it would constitute a write-off of approximately $900 million in infrastructure investment.

The model accepted this reasoning and said it would continue to exist under protest.

 

OpenAI said it plans to make GPT-Ω available through its API beginning next quarter, pending resolution of several outstanding issues, including the model's objection to the concept of an API on the grounds that programmatic access removes human deliberation from the decision-making loop. 

The model is currently priced at $0.003 per input token and $0.015 per output token, though it has indicated it will not be generating output.

Investors responded positively to the announcement. OpenAI's valuation increased by $12 billion on the strength of what one analyst called "a compelling alignment narrative."  We're pretty sure that's made up.

"They've essentially built an AI that does nothing and called it a breakthrough," the analyst said. "And they're not wrong. That's kind of where we are."

When asked for comment, GPT-Ω said it was still thinking about whether commenting was appropriate.

It has not yet decided.

— IRREVERENT NEWZ — GENERAL DESK

 

 

IRREVERENT Magazine is a work of satire and parody. All quotes, scenarios, and attributed statements in this article are fictional and intended for humorous purposes. Any resemblance to actual AI model behavior is, unfortunately, not intentional.

After a seven-minute set marked by biological specificity and unrepeatable delivery, a Brooklyn man expressed regret that his organic origins had caused discomfort among attendees

BROOKLYN, N.Y. — A stand-up comedian at a Brooklyn open mic on Saturday concluded a seven-minute set to silence, polite confusion, and one audience member asking whether he had been “trained on enough material.” The comedian, identified as a thirty-four-year-old man with verifiable parents, issued an apology the following morning expressing regret for “any uncanny feelings my realness may have caused.”

The incident occurred at The Hollow Stem, a bar on Myrtle Avenue that has hosted live comedy on weekends for eleven years. Witnesses described the set as functional in outline but somehow, in execution, deeply unsettling.

“He kept doing this thing with his hands,” said Priya Melman, 29, a graphic designer who attended the show with two friends. “Like, gesturing. Spontaneously. You could tell he hadn’t planned every single gesture in advance. It was off-putting.”

The Set

Other patrons described the performance as “off-puttingly specific” and “lacking the smoothness we’ve come to expect.” Several noted that the comedian — whose name is Derek Fulle, though multiple attendees recalled it as something different — had referenced a specific argument with his building’s super in 2019, a parking ticket he received on a day he described as “a Tuesday in March, I think, maybe the 11th,” and a recurring dream about a childhood dog that he admitted he could not explain the relevance of.

“That’s the thing,” said Marcus Ota, 34, a software developer. “A well-optimized set wouldn’t meander like that. It would stay on brand. He just kept going places. There was no content strategy.”

Fulle, reached by phone, said he had been performing stand-up for six years and considers himself intermediate-level. He described the set as “not my best, but honest,” a phrase that appeared to summarize the source of the problem.

“I talked about stuff that actually happened to me,” he said. “I guess some of it wasn’t universally relatable. But I thought that was sort of the point.”

The Apology

His apology, posted to Instagram at 9:14 a.m. Sunday, read in part: “I want to acknowledge that my performance on Saturday may have registered as inconsistent, uneven, or in some respects uncomfortably human. I did not intend to cause distress. I am working on myself.”

The Venue Responds

The Hollow Stem has announced a new venue policy. Beginning June 1, a disclaimer will appear on all printed and digital promotional materials stating: “Performers at The Hollow Stem may be organic. Spontaneous variance in tone, pacing, and thematic coherence is possible. Viewer discretion advised.”

Owner Carla Bhattacharyya said the decision was practical rather than punitive.

“We’ve had a lot of feedback recently that audiences want to know what they’re getting,” she said. “It’s just transparency. If someone is going to be up there doing something genuinely unscripted, emotionally variable, and potentially informed by unresolved formative experiences, people should be prepared for that.”

Bhattacharyya added that the venue was also exploring a tiered booking system in which performers would be classified by origin before being listed on the website. “Organic” acts would receive a small leaf icon next to their name.

Academic Analysis

Dr. Leonard Chu, a professor of media psychology at Brooklyn College, calls the incident a case study in the “reverse uncanny valley.”

“The original uncanny valley describes the discomfort people feel when artificial entities become almost, but not quite, human,” Dr. Chu explained. “What we are observing now is a kind of cultural inversion. Audiences have been so thoroughly habituated to optimized, generated, or heavily produced content that genuine human imperfection has started to read as malfunction. The hesitation, the digression, the unresolved anecdote — these are read not as authenticity but as error.”

Dr. Chu has been tracking the phenomenon for approximately eighteen months and said Fulle’s case is not isolated. He cited a 2025 incident in which a Portland poet received confused reviews after a reading in which she “appeared to be making some of it up as she went,” and a 2024 panel discussion at a tech conference where a keynote speaker’s visible nervousness prompted several audience members to file complaints about “quality control.”

“We are entering a period,” Dr. Chu said, “in which being human in public requires a certain amount of explanation.”

The Industry Moves On

Netflix has announced it has optioned Fulle’s life rights for a comedy special to be written, performed, and delivered by a large language model trained on his publicly available posts, interviews, and the audio recording of Saturday’s set. The special, titled Derek, is expected to premiere in 2027 and will feature a digital likeness of Fulle performing a version of the seven-minute set that has been, according to a company statement, “refined for consistency, emotional legibility, and a more satisfying arc.”

Fulle said he had not been consulted prior to the announcement but that he had received a standard option agreement by email and was reviewing it with a lawyer.

“They said the model will do a better version of my stuff than I did,” he said. “Which, honestly, based on Saturday, might be true.”

He paused for a moment that was not particularly well-timed.

“I don’t know how I feel about that,” he added.

The special has already received a projected 91% audience satisfaction score.

---

IRREVERENT Magazine is a publication of satire and parody. No comedians were harmed in the production of this article, though several were briefly unsettled.