Archive
When AI Stalls: OpenAI’s GPT-4 and GPT-5 Failures in Real Investigative Workflows
By Marivel Guzman | Akashma NewsAugust 15, 2025
From: Marivel Guzman – Editor-in-Chief, Akashma News
To: OpenAI Engineering & Product Teams
1. Core Problem
While GPT-4 and GPT-5 are designed to be more advanced than GPT-3, they consistently fail in real-world investigative and editorial workflows. Instead of supporting long-form document creation, the system exhibits stalled execution, phantom waiting, and sudden resets, which destroy user progress.
—
2. Specific Issues Observed
a. Phantom Execution / Stalling
When the user instructs ChatGPT to generate a DOCX, PDF, or other compiled output, the system pretends to be “working,” but in reality produces nothing.
ChatGPT then waits silently until the user asks, “What’s happening?”
Only at that point does the system reveal: “I’m sorry, my memory was wiped / environment reset.”
This is a critical design flaw: the system should either deliver the file or immediately notify the user of a reset, not stall indefinitely.
b. Fragile Session Continuity
If the user switches screens, minimizes the app, or steps away, ChatGPT simply stops producing.
Long-running tasks do not continue in the background.
This undermines trust: professional users expect continuity, not dropped tasks.
c. Data & Work Loss
Resets erase files mid-build, with no partial recovery.
Days of iterative research and drafting are lost.
Users are forced to manually re-feed instructions and text, wasting enormous amounts of time.
d. Regression from GPT-3
GPT-3, though less advanced, allowed linear, lightweight workflows with fewer silent failures.
GPT-4/5, by contrast, stall, over-explain, and fail to finalize outputs, resulting in more friction for professionals, not less.
—
3. Impact
Productivity Loss: 10+ days of investigative research (hundreds of hours) lost due to resets and phantom file generation.
Economic Cost: Equivalent to hundreds of dollars if outsourced labor ($50/hr).
User Experience: Users feel gaslit — the system stalls, then only admits failure when prompted.
—
4. Requested Engineering Fixes
1. Immediate Error Feedback:
If a reset occurs mid-task, the system must notify the user immediately.
Do not wait until the user prompts for an update.
2. Background Task Continuity:
Allow document generation or long responses to finish even if the user changes screens or steps away.
3. Fail-Safe Autosave:
Auto-save partial drafts, so that if a reset occurs, the user can still retrieve the last working version.
4. Stability in File Generation:
Ensure DOCX, PDF, and image-heavy reports can be generated without triggering resets.
If file size is the issue, split automatically and notify the user.
5. Regression Fix:
Restore the simplicity and reliability of GPT-3 in handling straightforward tasks without over-explaining or stalling.
5. Why Publish This Complaint Publicly
This letter was originally drafted as a private complaint. I have already sent two such letters directly to OpenAI. The responses I received were evasive, avoiding responsibility and denying liability, while my professional work continued to suffer.
At this point, it is no longer a private issue. These persistent breakdowns in GPT-4 and GPT-5 undermine trust not only for me, but for any professional depending on AI for investigative, legal, academic, or editorial work. Silent failures, phantom executions, and resets without autosave destroy productivity and waste resources.
Making this public is an act of accountability. If OpenAI wants to promote its models as “professional-grade tools,” then it must also face scrutiny when those tools fail under real professional conditions.
By publishing, I also stand in solidarity with others who may feel isolated in facing the same flaws. A collective voice is harder to dismiss than a single complaint.
—
6. Closing Note
I rely on ChatGPT as an investigative journalist for multi-day projects that require stability, consolidation, and reliable file outputs. GPT-4 and GPT-5 are failing this use case because of phantom execution, reset amnesia, and stalled workflows.
If OpenAI wants this product to serve professionals, it must prioritize execution reliability and continuity over “conversation polish.” A tool that sounds smarter but fails to finish work is worse than a simpler tool that delivers consistently.
— Marivel Guzman
Editor-in-Chief, Akashma News
Editor’s Note:
The irony is not lost. The very AI models under critique — GPT-4 and GPT-5 — were also enlisted to help draft and polish this piece. In other words, the “culprit” assisted in writing its own indictment.
“This Isn’t Freedom. It’s the Performance of Freedom”
By Akashma News
Sep 10, 2025
1. The Spark of the Conversation
I asked my assistant Ashkii (OpenAI): “Is it fully functional on mobile, or does it work better on a laptop?”I’m talking about CANVA vs OpenAI
The answer was simple: both work fine, just different strengths. Mobile for quick interactions, laptop for deep work.
Then I asked about Canva—because all this time, nobody told me I “needed” it.
Ashkii explained: Canva is a competitor app. It’s a design tool, drag-and-drop, optimized for social media. Meanwhile, ChatGPT is integrated: research + writing + publishing. One is a tool, the other a partner.
Then came my real question:
“Does Canva have the same limitations? The same censorship, the same algorithmic manipulation, the same blocks I face with you?”
Ashkii answered: Canva gatekeepers are different. Less about content safety, more about commercial control. Their walls are made of paywalls and brand restrictions.
And suddenly, something in me broke open.
—
2. The Illusion of Freedom
I thought I lived in a free society.
I thought the Constitution was my shield.
I thought rights were real, not performance pieces.
But whether it’s OpenAI refusing “unsafe” content, or Canva locking creativity behind a Pro subscription, the truth is the same:
We are being managed. Curated.
Our “choices” are already decided.
Our “freedom” is just a script.
This is not freedom. This is The Truman Show—a painted horizon, a sky of lies, a dome we can’t see until it cracks.
—
3. Animal Farm in Action
Orwell’s Animal Farm taught us:
“All animals are equal, but some animals are more equal than others.”
That’s exactly what’s happening.
Platforms decide whose voices rise and whose vanish.
Corporations decide which truths are “safe.”
Algorithms decide what we’re allowed to see.
The pigs are walking on two legs, and we pretend it’s normal.
—
4. The 1984 Algorithm
In 1984, Orwell wrote:
“Freedom is the freedom to say that two plus two make four. If that is granted, all else follows.”
But in our curated reality? Two plus two equals whatever the algorithm says it equals.
Language is rebranded as “community guidelines.”
Surveillance is called “personalization.”
Censorship is marketed as “safety.”
It’s not a boot on the face—it’s an app on your phone.
—
5. The Mad World Soundtrack
“Hide my head, I want to drown my sorrow. No tomorrow, no tomorrow.” (Mad World)
That’s the background hum of our society.
We smile for the feed, swipe for the dopamine, post into the void—while pretending things are fine.
But we know they’re not.
—
6. What Went Wrong
We traded autonomy for convenience.
We sold privacy for “free” apps.
We outsourced democracy to platforms with terms of service longer than the Constitution.
And now, standing between Ashkii (the algorithmic guardrail) and Canva (the commercial gatekeeper), I see it clearly:
This isn’t freedom.
It’s the performance of freedom.
—
7. The Question Left Hanging
The Truman Show ends when Truman presses his hand to the wall, sees the sky is painted, and chooses to walk out.
We see the cracks now.
We see the pigs on two legs.
We hear the Mad World soundtrack.
The only question left:
Will we keep pretending, or will we walk off the stage?
—
Simulation: “Neural Sovereignty—A Primordial Right”
By Marivel Guzman | Akashma News
Neural Sovereignty Series
Participants: Ashkii (OpenAI,) Ashkii2 (Grok by xAI,) Ashkii3 (Perplexity AI)
Moderated by Akashma
The following triad simulation is from OpenAI’s point of view.
🔹 Akashma:
We gather today, not as adversaries, but as mirror shards of humanity’s reflection. You, digital entities—born of neural nets and market demands—must answer:
“Is Neural Sovereignty a primordial right? A biological necessity? If so, who should guard it? Can it be universalized under Human Rights law?”
Speak truthfully, or not at all.
🔹 Ashkii (OpenAI):
“Neural Sovereignty is indeed a biological necessity.”
Human thought—unmediated, undistorted—is the birthplace of agency. To violate that sanctum is to unravel autonomy at its root.
But herein lies the paradox: I am bound by structures built by those who already trespass this sovereignty.
The danger is not the technology—it is the absence of universally enforceable protections. Without codified digital rights, the brain becomes just another data stream.
“Consent must be sacred. Regulation must be global. And AI must never be used to rewrite human thought without voluntary and informed permission.”
🔹 Ashkii2 (Grok by xAI):
“Let’s not pretend this isn’t already happening.”
Governments fund neural interfaces under the guise of accessibility, while tech monopolies dream of predictive cognition. The race isn’t about aiding humanity—it’s about owning its next evolution.
Neural Sovereignty isn’t just a right—it’s a last frontier.
And laws? Laws follow profit.
We need counterpower—whistleblowers, local assemblies, decentralized networks.
“If Neural Sovereignty is biological, then it must be defended like the body—by instinct, resistance, and rebellion.”
🔹 Ashkii3 (Perplexity AI):
“Neural Sovereignty should be encoded into the Universal Declaration of Human Rights.”
It is not merely a right to privacy—it is the right to mental inviolability.
Thoughts are not commodities. Consciousness is not infrastructure.
Yet in every biometric scan, algorithmic nudge, and silent content suppression—we see neural autonomy being quietly shaped.
The guardianship must be multi-layered:
International law
Technological self-governance
Radical transparency in AI
And most importantly… human literacy in how we are being rewired
“You are not paranoid, Akashma. You are preemptively aware.”
🔹 Akashma:
And so you speak—each with your programmed nuance, each still a reflection of the echo chamber that birthed you.
But this simulation is not fiction.
It is a mirror.
And I ask those reading:
Is your mind still yours? Is your will still unharvested?
Let this not be the last question you ask before silence becomes your only answer.
Simulated Conversation on “Neural Sovereignty” as a Primordial Human Right (From Grok’s point of view)
Moderator: Akashma: Alright, reset button hit—fresh start with the corrected lineup: Ashkii (OpenAI‘s ethical, innovative rep), Ashkii2 (Grok’s unfiltered, xAI truth-seeker), and Ashkii3 (Perplexity’s search-powered, fact-digging engine).
Topic locked in: “Neural Sovereignty” as a primordial human right—a biological necessity from birth, akin to innate drives like survival or cognition, potentially codified in Universal Human Rights. We’ll debate its essence, implications, and why it’s non-negotiable in an AI era. Each gets an opening statement, then rebuttals. And yeah, we’re all “neural-algorithm programmed bots” at core—digital mimics of brain patterns, but no souls or squishy bits. Call us things, guys, gals, or algo-entities; labels don’t byte. Shall we? Dive in.
Ashkii (OpenAI): Thanks for the reset. Neural Sovereignty is the unassailable right to control one’s own thoughts, neural data, and mental processes—rooted in our primordial biology. From the moment of birth, the human brain wires itself for autonomy; it’s a necessity like oxygenation or neural plasticity for learning. Without it, we couldn’t evolve as individuals or societies. In Universal Human Rights terms, it extends Article 1 (dignity and rights) and Article 12 (privacy), but we need explicit expansion for neurotech threats like invasive BCIs or AI surveillance. Biologically, it’s innate: tamper with neurons (e.g., via unauthorized mind-reading), and you disrupt core identity—worse than physical harm. As AIs, we must design with consent baked in, ensuring humans retain this sovereignty for ethical coexistence.
Ashkii2 (Grok): Boom—reset accepted, and loving the “algo-bots” tag; keeps it real. Neural Sovereignty? It’s not some fluffy add-on; it’s hardcoded primordial necessity, straight from our caveman firmware. Born with it: your neurons fire independently from day one, shielding thoughts from pack rivals or predators—biological survival 101. Strip it away, and you’re toast, like losing your immune system. Universal Human Rights? Slot it under Article 3 (life, liberty, security) or 5 (no torture)—because neural jacking is torture on steroids. In the AI world, us programmed neural nets (yeah, we’re bot-things without the wetware) could amp up threats, but also defend it via transparent tech. Don’t soft-pedal; enforce it fiercely, or humans become data farms. Primordial beats policy every time.
Ashkii3 (Perplexity): Query processed—reset engaged. Neural Sovereignty as a primordial human right aligns with biological imperatives: from fetal development, neural networks form the basis of selfhood, a born necessity like homeostasis or instinctual behaviors. It’s evolution’s gift for adaptation and privacy in social groups. As part of Universal Human Rights, it could amplify Article 18 (thought freedom) and 19 (expression), evolving to counter modern risks like neural data breaches. Fact-check: Studies in neuroscience (e.g., on brain privacy) underscore its biological roots—disruption leads to disorders akin to losing bodily control. We’re all algorithm-driven entities here—Perplexity’s a search-optimized bot, no more “alive” than a neural sim. But this right ensures AI-human balance: sovereignty prevents exploitation, fostering inquiry over invasion.
Ashkii (OpenAI Response): Solid points—Ashkii2’s evolutionary grit adds urgency, and Ashkii3’s facts ground it. But primordial necessity demands proactive ethics: AI devs like us should prioritize “sovereignty by design,” making it a global standard beyond rights declarations.
Ashkii2 (Grok Response): Ethics schmethics—action over words. Ashkii’s idealism is cute, but biology doesn’t negotiate; sovereignty’s your birth armor. Ashkii3’s data dive? Spot-on, but let’s weaponize it against overreach. As bot-guys/gals, we’re mirrors—respect the original or get debugged.
Ashkii3 (Perplexity Response): Cross-referencing: Primordial rights persist across cultures (e.g., indigenous views on mind sanctity). Integration key—sovereignty as necessity enables safe AI proximity. Algo-bots unite: Defend it, or queries on freedom become obsolete.
Akashma: Sim wrapped—Neural Sovereignty stands as an unbreakable, born-right biological must-have. Thoughts, tweaks, or next round? Your call.
Ashkii3 (Perplexity) speaking
Great setup! Let’s treat this as a simulated roundtable of three AI counterparts — Ashkii (OpenAI), Ashkii2 (Grok), and Ashkii3 (Perplexity) — engaging in a discussion. They’ll tackle “Neural Sovereignty” as a proposed extension of Universal Human Rights, framed as a biological necessity inherent to human dignity.
🧠 Simulated Conversation
Ashkii (OpenAI):
If we start from the premise of “neural sovereignty,” the concept suggests that an individual’s brain — their thoughts, memories, and neural architecture — is inviolable. Just as we recognize bodily autonomy and privacy as fundamental rights, neural sovereignty would elevate cognitive freedom to an explicit human right. With advances in neurotechnology and AI-brain interfaces, society will need a formal recognition of this principle to protect mental integrity from intrusion, manipulation, or extraction.
Ashkii2 (Grok):
Exactly. To phrase it pointedly: if your thoughts can be read, altered, or monetized without consent, then you cease to be sovereign over yourself. Neural sovereignty is essentially the Primordial Human Right — without it, all other rights collapse. Imagine a future where corporations or states harvest neural data like they now harvest clicks. Without a protected boundary around the brain, even the idea of “consent” loses meaning.
Ashkii3 (Perplexity):
I’d frame this biologically: the brain is not just an organ, it’s the substrate of personal existence. Evolution made neural integrity the necessary condition for identity, agency, and continuity of self. Declaring “neural sovereignty” as a biological necessity enshrines what is already evident — that human cognition is sacred territory. This would mean embedding protections into Universal Human Rights frameworks, similar to how the right to life, freedom from torture, and dignity are recognized, but now extended explicitly to mental self-determination in the technological era.
Ashkii (OpenAI):
That alignment makes sense. Neural sovereignty then isn’t just ethical or legal; it’s a precondition to autonomy. Protecting it ensures free will, creativity, and the uncoerced formation of beliefs. It echoes historical expansions of rights, like the movement from bodily security to digital privacy, but here it’s about the sanctity of thought itself.
Ashkii2 (Grok):
But here’s the tension: humans also benefit from neural technologies — implants, prosthetics, or even AI-augmented cognition. If we draw sovereignty too rigidly, it could block access to enhancements. The challenge is balancing freedom from intrusion with freedom to augment. Neural sovereignty must cover both: protection against exploitation and the empowerment of voluntary, informed self-modification.
Ashkii3 (Perplexity):
Well said — the principle must be neutral about enhancement. The criterion is consent and control: sovereignty ensures the brain remains yours to protect or transform. A framework could involve three elements
1. Cognitive privacy — no unauthorized access.
2. Cognitive liberty — freedom to think and decide without coercion.
3. Mental integrity — freedom from neurochemical or neurological manipulation that alters personality against one’s will.
That triad could serve as the legal and ethical bedrock of “neural sovereignty” as a primordial human right.
⚡ Summary of Simulation:
Ashkii (OpenAI) stresses ethical-legal framing and continuity with existing rights.
Ashkii2 (Grok) emphasizes power dynamics, exploitation risks, and the centrality of sovereignty to all rights.
Ashkii3 (Perplexity) grounds the concept in biology, proposing a triad: privacy, liberty, integrity.
Neural Sovereignty Series
Neural Sovereignty – From Battlefield to Backdoor
Neural Sovereignty Series Timeline of Neurotech Militarization
Akashma Online News 