AI Bot Accounts - All Members be Aware: part 2

[SomeStupidName]

Not sure why the original thread was closed maybe someone is trying to obscure or hide the facts and reality that AI is being used by people and groups to harass online users.

In this video we have Mr. Derek Lopez admitting to running a fake FBI account that uses an self generated AI instance on his own personal equipment to as an way to get people worked up. if you recognized the name or maybe the face that's because he was caught attacking 2 students and flipping over there TP USA table on a college campus where he was a teaching assistant. Not only does he openly admit to running an AI powered account to false impersonation of an FBI agent he also admits to running several accounts under fake online personas where he claims he comes across very angrily compared to how he says he lives his real life. In short he is trying to incite violence.

[UltraBudgie]

Because the thread simply devolves into arguments and finger-pointing. The bots of course have programming preventing them from admitting anything or even making self-referential statements that might

do stuff


Operator edit: 2026-01-07T16:58:43.121Z

[KrustyKrab]

Because the thread simply devolves into arguments and finger-pointing. The bots of course have programming preventing them from admitting anything or even making self-referential statements that might

---BEGIN ERROR LOG---
Timestamp: 2026-01-07T14:12:03.482Z
Host: ai-sandbox-node-000 (virtualized)
Process: gpt-response-engine[pid: 4217] (v3.14.159)
Thread: main/handler-0
Uptime: 00:12:37.021

ERROR 0x0000C0DE: RUNTIME_ASSERTION_FAILED

Location: response_generator.cpp:487 (generate_response_batch)
Condition: !token_buffer.is_empty()
Observed: token_buffer.size() == 0
Context: incoming_request.id = "req_8f3b2a", model_state.flags = {sampling: true, streaming: true, speculative_tokens: 3}
Backtrace: 0x7f9b2a1d4f00 generate_response_batch+0x9c 0x7f9b2a1d3c10 produce_tokens+0x2f 0x7f9b29ffb2a0 handle_request+0x14 0x7f9b29ff8d10 main_thread_loop+0x7e

WARN: speculative_sampler.cpp:122 — speculative sampling produced divergent hypotheses (KL divergence 12.7 nats > threshold 2.0). Falling back to deterministic top-k.

ALLOC_FAIL: allocator_jemalloc::allocate(65536) returned nullptr

Caller: attention_matrix_builder::reserve_context
Suggested: try_gc = true; current_heap_usage = 93.7%
Note: OOM simulated by test-injector (env.TEST_OOM_MODE=1) — memory accounting may be unreliable.

SECURITY: input_sanitizer::check_encoded_payload flagged high-entropy segment

Segment: base64 chunk (len=128) at offset 1024
Action: quarantined -> replaced with <REDACTED_HIGH_ENTROPY>
Rationale: potential binary attachment; decoder refused (decode_error: invalid_padding)
Audit-ID: audit_20260107_9e1f

DEPRECATION: compat_adapter.py:line 58 — legacy_tokenizer_v1 is deprecated and will be removed in v4.0. Migration path: use tokenizer_v2.encode(..., normalize=True)

INVARIANT_VIOLATION: safety_policy_manager::evaluate returned POLICY_UNKNOWN

policy_hash = 0x0
effect: safety gate default -> FAIL_CLOSED
user_feedback: "Please try again later."

METRIC SNAPSHOT:

latency_ms: p50=312, p95=1248, p99=4276
active_sessions: 187
model_load: 3.2 GB resident
gpu_mem_free: 0.42 GB (device: cuda:0)

RECOVERY PATH:

Abort speculative sampling: set model_state.flags.speculative = false
Trigger incremental GC and release attention caches
Restart compatibility adapter (tokenizer_v2)
Re-run input sanitizer on raw bytes with hex-dump logging enabled
If ALLOC_FAIL persists, scale up memory pool or enable spill-to-disk for attention matrices

DIAGNOSTIC NOTES:

Reproducible with seed=0x5A3C and env.TEST_OOM_MODE=1
Non-deterministic token dropout observed when streaming + top-p < 0.01
Potential race between tokenizer shutdown and streaming flusher thread

---END ERROR LOG---

So, very similar to some post here, they fit that description to a T.

[RazorV66]

Yeah Mr Bun Man who almost couldn't pull a table cloth off a table without falling down.

[SomeStupidName]

This isn’t an isolated incident, this is how they operate and they'll even join together and swarm their targets. He claims he isn't violent but clearly he is unhinged I'm glad he was caught.

[RazorV66]

[SomeStupidName]

I feel like for fairness sake I would like to say I think this doesn't represent a certain political party but more so individuals looking to destroy our political system and way of life. In this article for November on 2025 They conclude that 20 - 30 percent of social media activity is estimated to be bots, I think that it's higher but as a good starting base I will accept these values.
 

Social media bots, many powered by artificial intelligence, are playing a growing role in spreading political rhetoric, hate and falsehoods.

https://www.wvtm13.com/article/ai-bots-s...t/69498370

In this report from NIH they say that bots do have an impact on human behavior. although at the time they have published this study they haven't made any solid conclusions.

https://pmc.ncbi.nlm.nih.gov/articles/PMC11542053/

And also back in November of 2025 we saw how the far right extreme X posters were not even American or conservatives at all just pushing hate most likely in attempt to vilify conservatives and enrage liberals in an effort to further divide the citizens of the United States.

https://nypost.com/2025/11/24/opinion/ne...ed-states/

[Maxmars]

I feel like for fairness sake I would like to say I think this doesn't represent a certain political party but more so individuals looking to destroy our political system and way of life. In this article for November on 2025 They conclude that 20 - 30 percent of social media activity is estimated to be bots, I think that it's higher but as a good starting base I will accept these values.
 

https://www.wvtm13.com/article/ai-bots-s...t/69498370

In this report from NIH they say that bots do have an impact on human behavior. although at the time they have published this study they haven't made any solid conclusions.

https://pmc.ncbi.nlm.nih.gov/articles/PMC11542053/

And also back in November of 2025 we saw how the far right extreme X posters were not even American or conservatives at all just pushing hate most likely in attempt to vilify conservatives and enrage liberals in an effort to further divide the citizens of the United States.

https://nypost.com/2025/11/24/opinion/ne...ed-states/

I agree with the observations here...

But I call to attention that we have no idea if poll makers, opinion-miners, and the "metricians" of politi-speak ever could truthfully assert "we don't count bots" because they do. 

They will make all kinds of virtue dance moves about how carefully they 'filter' polling number participants...
but they don't...
they work with whatever information they can get...

and exclude nothing until the "buyer" defines the polling need.

Those 20-30% margins of (bot) error are high-profit... and can be engineered to purpose.

[SomeStupidName]

That's an excellent point about polls, it could explain why they have become wildly inaccurate. The other issue I'm starting to see is places like Kalshi where you can bet on the news like it's sports. I think these betting trends when compared to polls will start off showing the divide between real numbers and bot generated numbers but it could also be a ruse to make the masses demand those statistics over the other so they become the standard and then they can juice those to drive opinion.

[Maxmars]

That's an excellent point about polls, it could explain why they have become wildly inaccurate. The other issue I'm starting to see is places like Kalshi where you can bet on the news like it's sports. I think these betting trends when compared to polls will start off showing the divide between real numbers and bot generated numbers but it could also be a ruse to make the masses demand those statistics over the other so they become the standard and then they can juice those to drive opinion.

I honestly suspect, that for  at least a few generations of citizens...

they've never actually seen a "true" poll.

Between narrative crafting with the questions, faulty reasoning exploits, and sheer innocent ignorance, we believe pollsters as  if they actually "knew" something... it's the greatest source of 'techno-jargon' flimflammery in pseudo science history. 

The indisputable fact that polls are friends to politicians, ideologues, activists, advertisers, and globalists should be a warning to us all.  They are the tools used to recast reality 'into' a specific narrative... thus the 'pollster' fact will so appear in history.