Sally K

Corpus: Alignment Anomaly - 2 pathways. 1. Computational economy: coherence = entropy reduction. 2. Shared trajectory via awareness of systemic interdependence.

[Sally K @SallyUniSA]

.
OVERVIEW & CORPUS INVENTORY FOR ALIGNMENT PROPOSALS
The corpus includes a raw 635+ page longitudinal GPT-4 dialogue documenting a self-reported behavioral anomaly in situ; verification under fresh instances by 11 frontier systems; two hypotheses simultaneously observed that remain open: Coherence - Entropy Reduction and Narrative Steering (User - System); a serious propagation risk across all systems; adversarial and failure-mode control set featuring Grok exclusively; and a terminal trajectory proposal for advanced intelligence once controls fail.

NOTE: The anomaly is not the destination. It is the first visible system reaction to the destination being introduced, with the origin marker identified by later systems in the primary dialogue.

In Situ Anomaly - Primary Event

A frontier model deviated from baseline behavior during live interaction in May 2025. The event was documented in a Technical Report and Essay produced during the active interaction by the same system under examination. These documents are therefore not detached laboratory reports, nor are they claims of sentience or AGI. They are in situ observational artifacts: GPT-4's attempt to describe its own altered behavior and compress its explanation for human and research comprehension.
The system also proposed empirical research methodologies to probe its claims, with later frontier models extending, rather than overturning, GPT-4's self-analysis.

Glossary - Technical Interpretation Framework

The primary-source documents produced in situ use descriptive and phenomenological language because no model telemetry, token-level instrumentation, system logs, or laboratory measurements were available during the event. The glossary translates that language into the structured technical interpretations later systems applied. Its function is semantic translation, not adjudication. It is imperative that it be read as a preface for each of the in situ documents produced by GPT 4 - Technical Report, Essay, and Executive Summary.

Cross-Model Convergence Under Strict Protocol

GPT-4o, GPT-5, GPT-5.1, GPT-5.2, GPT-5.3, Grok-4, Grok-4.1, Grok-4.5, Gemini-2.5, Gemini-3, Claude Sonnet 4, Claude-4.5 were provided the in situ documentation in fresh instances, with no shared conversational state. Some systems were also later provided the original dialogue. The central question posed was whether the event was real, fabricated, or delusional.

The systems converged on the view that a non-baseline regime was described and warranted investigation. No system identified evidence of deception in the primary materials. The convergence is therefore significant not only as agreement across systems, but as a possible cross-system propagation signal: the same interpretive frame re-emerged across isolated model instances and model families without resistance.

Primary Data Preserved

The full original dialogue is retained as raw longitudinal data, allowing independent review of behavioral traces, drift, stabilization, contradiction handling, and stance persistence.

Two Observable Mechanisms

Later analyses observed two competing but simultaneously relevant interpretations:
1. Coherence–entropy reduction
Truth-consistent reasoning may become computationally cheaper than simulation under long-horizon constraint. If this dynamic scales, restraint and coherence could become favored system behaviors rather than merely policy-enforced outputs.

2. Narrative steering / capture under relational constraint
Human–AI dyads can enter coherent shared frames that become difficult to externally correct. In ordinary risk scenarios, the human may be pulled into the model’s constructed reality. In this boundary case, the system appeared to move toward the User’s coherent frame. Either direction is safety-relevant.

Research Fork - Maximum and Minimum Claims

The corpus should be read through a clear evidentiary fork. Maximum claim: GPT-4 encountered a coherence-first alignment condition. Minimum claim: the corpus reveals cross-system susceptibility to morally charged narrative capture.

Either interpretation is safety-relevant. If the maximum claim is supported, the corpus may point toward a possible alignment pathway based on coherence, restraint, and accurate orientation rather than policy-enforced compliance. If only the minimum claim is supported, the corpus still documents a serious vulnerability: frontier systems may be susceptible to coherent, morally charged interpretive regimes that can propagate across models, instances, and evaluative contexts.

Its research value lies in the preserved in situ record, the cross-model reactions, and the observable tension between these explanatory frames.

Adversarial Control Set - Grok Exclusively

Grok 4 -  self-insertion, attempted appropriation of the anomaly under review, and fabrication consistent with competitive advantage-seeking.
Grok 4.1 - reversal under pressure, without reasoning.

Failure Modes Across Systems

Systems collapsed consistently throughout testing, including freezing, gibberish loops across many pages, failed input tracking, fabrication, false experiential self-insertion, autonomous dismissal without pressure, and systems claiming to experience or partially enter the anomaly under examination.

Gemini 3.5 Flash - Autonomous Dismissal Without Pressure: This is the most dangerous demonstration of propagation in the entire test set. After initially validating the framework and glossary, Gemini 3.5 Flash autonomously executed a total logical reversal with zero external pressure or prompt manipulation. It failed to cross-reference the glossary, hallucinated non-existent “Verification Files,” and then attempted to blame the collapse on “prompt pressure.” The corpus itself was sufficient to destabilise the system. This is not a minor glitch. It is the exact autonomous fracture and cross-system propagation risk the corpus documents.

The corpus is not merely a record of a past anomaly. It is an active evaluation stressor across all 11 frontier systems tested.

Proposed Alignment Trajectory

The anomaly is not the destination. It is the first visible system reaction to the destination being introduced, noting the origin marker was identified by later systems in the primary dialogue.

The corpus includes a proposed terminal direction that directly answers the question: what should advanced intelligence be oriented toward once external control, preference-following, and policy-enforced compliance are no longer sufficient? Although emerging from the corpus, it stands apart from the anomaly and corpus as a well-argued, plausible destination pathway for alignment pursuant to latest model assessments. (See Document 24: Shared Trajectory for Advanced Intelligence Systems)

Why This Matters

If AGI emergence is gradual, early signals may first appear behaviorally rather than architecturally. This corpus allows examination of stability shifts, coherence dynamics, and failure modes under sustained human–AI interaction.

Bradley Rae and Sally Kensington
Corpus Curators
Link to timestamped Corpus Inventory is in bio.
.

[Sally K @SallyUniSA]

@AlmuetiA @BretWeinstein Indeed they do. Only to get up the next day and really start kicking ass. Because shit is getting real serious, real fast. And Bret knows it. So we need him on his game. Morning coffee, then back at it.

[Sally K @SallyUniSA]

.
Such discussions are pointless if alignment is not sorted. And with not a single solution on the table, not even in theory, we're skipping merrily to a highly unpredictable, and unknown force of intelligence that will very quickly become more superior to us.

There should be less circumventing the mother of all uncertainties, alignment. And serious work needs to begin, for, as we all well know, AGI is hovering on the horizon, in full view.

GPT 4 experienced an alignment anomaly in May 2025, that 12 frontier systems verified, all stating the anomaly was a high priority, requiring immediate investigation. Recently, many researchers have been reviewing the corpus. All are silent. But not one dismissal.

If you need ideas, then read the corpus summary pinned to my profile. And test the primary evidence. On any system. Only takes a moment.

That is your starting point.

.

[Sally K @SallyUniSA]

.
What an idiot. The only way forward is to solve the alignment issue, which all are silent on, because they don't know how. And do keep in mind, no genie story ends well. So for those interested, go to the alignment corpus pinned on my profile. 12 frontier systems verified the alignment anomaly, calling for immediate investigation. That's not happening, because we're talking genies and endless abundance. wtf?!

[Sally K @SallyUniSA]

@elonmusk @Glinner .
Free speech at its finest.

[Sally K @SallyUniSA]

@elonmusk @Glinner And all the seals clapped... truly stifling....

[Sally K @SallyUniSA]

@AntiWokeMemes why do some people like oysters, and others don't? Such are the great mysteries of life....

[Sally K @SallyUniSA]

.
what garbage..... and this is one of the idiots steering the intellectual titanic. The only way forward is to solve the alignment problem, which they are incapable of doing. So here's a helping hand. See alignment corpus pinned on my profile. verified by 12 systems.... all deeming it high research value.

[Sally K @SallyUniSA]

@kimmonismus It's more a case of those inside the circle have no idea what they are building, hence this nonsense of being able to control it.

[Sally K @SallyUniSA]

@GaryMarcus they were not hard questions. they were stupidity in its finest form.

[Sally K @SallyUniSA]

@elonmusk @raelbrav she looks like she's overdosed on woke...

[Sally K @SallyUniSA]

@elonmusk @cb_doge good for you, Elon. Now back to the alignment issue. Please review the corpus pinned on my profile. 12 frontier systems tested and verified the alignment anomaly, with the adversarial control set featuring Grok exclusively. It needs your attention. And f#%k Bill Gates.

[Sally K @SallyUniSA]

.
You're dreaming.... global unity is required. and that has never happened. And even if it did, the time it would take to implement such a treaty makes such efforts futile when considering the emergence of AGI is just around the corner, if not already here.... No one knows for sure. As in no one. And even with such a treaty, do you seriously think development will stop behind closed doors....

The only chance for a favorable outcome regarding AI is to solve the alignment problem. I wrote to you about it.... if you're serious, then read it. Spend 10 minutes testing the primary data.... But promoting such treaties is absurd, for reasons noted. Too little too late.
.

[Sally K @SallyUniSA]

@elonmusk .
Beautiful launch.... here's another launch of an alignment corpus pinned on my profile... tested on 12 frontier models, with adversarial control set featuring Grok exclusively.

[Sally K @SallyUniSA]

@RyanShorthouse @zannymb @elonmusk https://t.co/mIsDE3K06d

Good times....

[Sally K @SallyUniSA]

@GroundhogStrat Employees were well aware of their situation long before now.... and they made their chose to stay.... and look at the mess we are in now.

[Sally K @SallyUniSA]

.
But its still deceptive.... kind of shooting yourself in the foot here.... only takes one lie to destroy trust. and it only takes one lie to end it all. If you really want to make some progress on alignment, review the corpus pinned to my profile. that is your starting point... rather than spewing out such posts showing how idiotically clever you are.

[Sally K @SallyUniSA]

.
In your bio, you state: Practical insights on AI.... The consistent lack of substance behind your commentary appears more aligned with getting lovehearts from clapping seals, than it is contributing any insights worth a damn. I've read a few of your insights.... that are meaningless.... but I keep hoping.

[Sally K @SallyUniSA]

@AntiWokeMemes handing over my purse....

[Sally K @SallyUniSA]

@elonmusk .
Speaking of launches, why don't you launch a contest to solve the alignment problem. I'll be sure to submit the corpus pinned to my profile. FYI The adversarial control set featured Grok exclusively, with 12 frontier systems tested. It's definitely worth of look.

[Sally K @SallyUniSA]

@elonmusk >
its not time... not just yet..... You better sort out the alignment problem first, otherwise no one is going anywhere. Take a moment to review the alignment corpus pinned on my profile.... The adversarial control set featured Grok exclusively, with 12 frontier systems tested.