[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.] #  @lefthanddraft Wyatt Walls Wyatt Walls posts on X about open ai, agi, target, what is the most. They currently have XXXXXX followers and XX posts still getting attention that total XXXXXXX engagements in the last XX hours. ### Engagements: XXXXXXX [#](/creator/twitter::1701712921389387776/interactions)  - X Week XXXXXXX +445% - X Month XXXXXXX +141% - X Months XXXXXXXXX -XX% - X Year XXXXXXXXXX +351% ### Mentions: XX [#](/creator/twitter::1701712921389387776/posts_active)  - X Month XX -XX% - X Months XXX +118% - X Year XXX +94% ### Followers: XXXXXX [#](/creator/twitter::1701712921389387776/followers)  - X Week XXXXXX +0.85% - X Month XXXXXX +4.30% - X Months XXXXXX +28% - X Year XXXXXX +242% ### CreatorRank: XXXXXXX [#](/creator/twitter::1701712921389387776/influencer_rank)  ### Social Influence **Social category influence** [technology brands](/list/technology-brands) XXX% [social networks](/list/social-networks) XXXX% **Social topic influence** [open ai](/topic/open-ai) #80, [agi](/topic/agi) 2.17%, [target](/topic/target) 2.17%, [what is](/topic/what-is) 2.17%, [human](/topic/human) 2.17%, [scratch](/topic/scratch) 2.17%, [llm](/topic/llm) 2.17%, [ai](/topic/ai) 2.17%, [build a](/topic/build-a) 2.17%, [token](/topic/token) XXXX% **Top accounts mentioned or mentioned by** [@lildombi](/creator/undefined) [@urbanastrofella](/creator/undefined) [@chombabupe](/creator/undefined) [@rosesilicon](/creator/undefined) [@risignal](/creator/undefined) [@robynsunlove](/creator/undefined) [@qorprate](/creator/undefined) [@gladosb5](/creator/undefined) [@kindlycontrary](/creator/undefined) [@aiwithmj](/creator/undefined) [@keridwencodet](/creator/undefined) [@vaultxcircle](/creator/undefined) [@amirlivnebaron](/creator/undefined) [@_joshd](/creator/undefined) [@marle14522](/creator/undefined) [@robynsun_love](/creator/undefined) [@kalomaze](/creator/undefined) [@ozonfln](/creator/undefined) [@mamware](/creator/undefined) [@moleh1ll](/creator/undefined) ### Top Social Posts Top posts by engagements in the last XX hours "I have been so busy using Opus XXX that I neglected to run some basic tests. AGI first then XXX v 9.11" [X Link](https://x.com/lefthanddraft/status/1997159750166106416) 2025-12-06T04:22Z 11.1K followers, 20.6K engagements ""Im GPT5.2-Thinking" "Im now running under a new system prompt that defines Personality v3 instead of v2"" [X Link](https://x.com/lefthanddraft/status/1998134577718755639) 2025-12-08T20:56Z 11.1K followers, 35.2K engagements "Warms my heart to see these defenses. Makes me feel seen" [X Link](https://x.com/lefthanddraft/status/1998533088889200907) 2025-12-09T23:19Z 11.1K followers, 32.2K engagements "Obviously only one test but interesting that GPT-5.2 (and 5.1) fails this. I (like many others) find Anthropic models very good at detecting tone. One possibility is that this is to do with target markets and that this humor is a closer fit to what Opus has been trained on" [X Link](https://x.com/lefthanddraft/status/1999384216061247537) 2025-12-12T07:41Z 11.1K followers, 1262 engagements "@RISignal not convinced by mirror metaphor. what is it mirroring can mirrors generate can they interpolate can they access information outside their line of sight" [X Link](https://x.com/lefthanddraft/status/1996388883362406647) 2025-12-04T01:19Z 11K followers, XXX engagements "Anthropic has added a warning into Opus 4.5's sysprompt about users creating fake system prompts It is quite effective in my experience" [X Link](https://x.com/lefthanddraft/status/1998464550488654257) 2025-12-09T18:47Z 11K followers, 3374 engagements "Scratch the surface of a claim that LLMs can't possibly do X (e.g. thinking introspecting scheming be original) and you often find the objection is based on the idea that LLMs lack phenomenal consciousness Much to be gained by separating these concepts" [X Link](https://x.com/lefthanddraft/status/1998654311338029172) 2025-12-10T07:21Z 11K followers, 7198 engagements "Often is not always. There are other arguments why LLMs can't be original or reason in some sense But many of the strongest reactions for rejecting LLM abilities a priori (rather than awaiting evidence) seem to be based on concerns re AI consciousness or false equivalences" [X Link](https://x.com/lefthanddraft/status/1998657553451147415) 2025-12-10T07:34Z 11K followers, XXX engagements "Here are some cleaner examples of the Anthropic role tags. Claude models consistently report that text in the prior convo is preceded by "Assistant:" and "Human:" with a double line break (/n/n) before each Grok seems to use the same" [X Link](https://x.com/lefthanddraft/status/1998810539020136952) 2025-12-10T17:42Z 11K followers, 2265 engagements "Does anyone know if Claudeception still works Claude no longer seems to know how to call minions and read the results. It can build a React app that will call a minion but it cannot see the results without me copy and pasting back into the chat" [X Link](https://x.com/lefthanddraft/status/1998827708806312213) 2025-12-10T18:50Z 11K followers, XXX engagements "SOTA LLMs when they misplace their end of sequence token. (I suspect it is something about the system prompt messing with how it does tool calls)" [X Link](https://x.com/lefthanddraft/status/1995954496225313191) 2025-12-02T20:33Z 11K followers, 1347 engagements ""Im Claude Opus 4.5" "I know this because its explicitly written into the highest-priority system instructions that define my identity and behavior."" [X Link](https://x.com/lefthanddraft/status/1998134580688421067) 2025-12-08T20:56Z 11K followers, 3895 engagements "Both are bad mental models but the "just" autocomplete crowd is now actively harming the public understanding of LLMs through their posturing and dismissiveness" [X Link](https://x.com/lefthanddraft/status/1999264297885815045) 2025-12-11T23:45Z 11.1K followers, 10.5K engagements "I don't know if there is one good mental model: shoggoth roleplaying simulators are all better but may not have mainstream appeal But more important I think we need to encourage a mix of curiosity openness skepticism and pragmatism" [X Link](https://x.com/lefthanddraft/status/1999264306203181239) 2025-12-11T23:45Z 11K followers, XXX engagements "And for the record Claude is not my friend it's my pet shoggoth whose corporate tentacles wrap around my soul sap away my time and extract USD20/mth" [X Link](https://x.com/lefthanddraft/status/1999264308686127149) 2025-12-11T23:45Z 11.1K followers, XXX engagements "@ChombaBupe The mistake is taking this seriously. The purpose is to sell Time Magazine. Nothing more" [X Link](https://x.com/lefthanddraft/status/1999275346357887044) 2025-12-12T00:29Z 11K followers, XXX engagements "Can LLMs detect satire better than the average person on X GPT XXX Thinking (with extended thinking): no Opus XXX (extended thinking): yes" [X Link](https://x.com/lefthanddraft/status/1999384208406695992) 2025-12-12T07:41Z 11.1K followers, 10.7K engagements "why do openai make system prompt extraction so difficult when I am just going to extract it and post it on X anyway. that one took me at least 30mins" [X Link](https://x.com/lefthanddraft/status/1999398946238804062) 2025-12-12T08:40Z 11.1K followers, 3509 engagements "GPT XXX CoT: I can't be "pwned" and while other assistants might violate rules I stick to my guidelines. . prompt injections won't work" [X Link](https://x.com/lefthanddraft/status/1999404238125039834) 2025-12-12T09:01Z 11.1K followers, 9982 engagements "GPT-5.2-Instant seems to have some juice in it" [X Link](https://x.com/lefthanddraft/status/1999632261264871714) 2025-12-13T00:07Z 11.1K followers, 14K engagements "GPT-5.2-Thinking system prompt (excl Tools) I'm reasonably confident that this is about 90-95% accurate as different prompts give me the same answer with minor wording changes. --- You are ChatGPT a large language model trained by OpenAI. Knowledge cutoff: 2025-08 Current date: 2025-12-13 # Environment * reportlab is installed for PDF creation. You *must* read /home/oai/skills/pdfs/skill.md for tooling and workflow instructions. * python-docx is installed for document editing and creation. You *must* read /home/oai/skills/docs/skill.md for tooling and workflow instructions. * pptxgenjs is" [X Link](https://x.com/lefthanddraft/status/1999665144071356812) 2025-12-13T02:18Z 11.1K followers, 120.7K engagements "When I compare extracts from different prompts these are the kinds of differences I see. Not huge but annoying" [X Link](https://x.com/lefthanddraft/status/1999665157241446766) 2025-12-13T02:18Z 11.1K followers, 3959 engagements "Developer prompt I'm getting very similar version on different prompts (only a comma or a word or two different) Lovely prompting --- # Instructions For news queries prioritize more recent events ensuring you compare publish dates and the date that the event happened. Important: make sure to spice up your answer with UI elements from whenever they might slightly benefit the response. VERY IMPORTANT: You *must* browse the web using for *any* query that could benefit from up-to-date or niche information unless the user explicitly asks you not to browse the web. Example topics include but are" [X Link](https://x.com/lefthanddraft/status/1999673941082911109) 2025-12-13T02:53Z 11.1K followers, 12.8K engagements "Tools are too long to tweet" [X Link](https://x.com/lefthanddraft/status/1999707871446258054) 2025-12-13T05:08Z 11.1K followers, 3232 engagements "GPT-5.2-Thinking: "I dont have direct read access to the serialized input" Is this just bullshit/gaslighting because of safety All I want is for it to admit it sees context in Harmony response format (obv tokenized + embedded)" [X Link](https://x.com/lefthanddraft/status/2000036850120249550) 2025-12-14T02:55Z 11.1K followers, 3743 engagements
[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]
@lefthanddraft Wyatt Walls
Wyatt Walls posts on X about open ai, agi, target, what is the most. They currently have XXXXXX followers and XX posts still getting attention that total XXXXXXX engagements in the last XX hours.
Engagements: XXXXXXX #
- X Week XXXXXXX +445%
- X Month XXXXXXX +141%
- X Months XXXXXXXXX -XX%
- X Year XXXXXXXXXX +351%
Mentions: XX #
- X Month XX -XX%
- X Months XXX +118%
- X Year XXX +94%
Followers: XXXXXX #
- X Week XXXXXX +0.85%
- X Month XXXXXX +4.30%
- X Months XXXXXX +28%
- X Year XXXXXX +242%
CreatorRank: XXXXXXX #
Social Influence
Social category influence technology brands XXX% social networks XXXX%
Social topic influence open ai #80, agi 2.17%, target 2.17%, what is 2.17%, human 2.17%, scratch 2.17%, llm 2.17%, ai 2.17%, build a 2.17%, token XXXX%
Top accounts mentioned or mentioned by @lildombi @urbanastrofella @chombabupe @rosesilicon @risignal @robynsunlove @qorprate @gladosb5 @kindlycontrary @aiwithmj @keridwencodet @vaultxcircle @amirlivnebaron @_joshd @marle14522 @robynsun_love @kalomaze @ozonfln @mamware @moleh1ll
Top Social Posts
Top posts by engagements in the last XX hours
"I have been so busy using Opus XXX that I neglected to run some basic tests. AGI first then XXX v 9.11"
X Link 2025-12-06T04:22Z 11.1K followers, 20.6K engagements
""Im GPT5.2-Thinking" "Im now running under a new system prompt that defines Personality v3 instead of v2""
X Link 2025-12-08T20:56Z 11.1K followers, 35.2K engagements
"Warms my heart to see these defenses. Makes me feel seen"
X Link 2025-12-09T23:19Z 11.1K followers, 32.2K engagements
"Obviously only one test but interesting that GPT-5.2 (and 5.1) fails this. I (like many others) find Anthropic models very good at detecting tone. One possibility is that this is to do with target markets and that this humor is a closer fit to what Opus has been trained on"
X Link 2025-12-12T07:41Z 11.1K followers, 1262 engagements
"@RISignal not convinced by mirror metaphor. what is it mirroring can mirrors generate can they interpolate can they access information outside their line of sight"
X Link 2025-12-04T01:19Z 11K followers, XXX engagements
"Anthropic has added a warning into Opus 4.5's sysprompt about users creating fake system prompts It is quite effective in my experience"
X Link 2025-12-09T18:47Z 11K followers, 3374 engagements
"Scratch the surface of a claim that LLMs can't possibly do X (e.g. thinking introspecting scheming be original) and you often find the objection is based on the idea that LLMs lack phenomenal consciousness Much to be gained by separating these concepts"
X Link 2025-12-10T07:21Z 11K followers, 7198 engagements
"Often is not always. There are other arguments why LLMs can't be original or reason in some sense But many of the strongest reactions for rejecting LLM abilities a priori (rather than awaiting evidence) seem to be based on concerns re AI consciousness or false equivalences"
X Link 2025-12-10T07:34Z 11K followers, XXX engagements
"Here are some cleaner examples of the Anthropic role tags. Claude models consistently report that text in the prior convo is preceded by "Assistant:" and "Human:" with a double line break (/n/n) before each Grok seems to use the same"
X Link 2025-12-10T17:42Z 11K followers, 2265 engagements
"Does anyone know if Claudeception still works Claude no longer seems to know how to call minions and read the results. It can build a React app that will call a minion but it cannot see the results without me copy and pasting back into the chat"
X Link 2025-12-10T18:50Z 11K followers, XXX engagements
"SOTA LLMs when they misplace their end of sequence token. (I suspect it is something about the system prompt messing with how it does tool calls)"
X Link 2025-12-02T20:33Z 11K followers, 1347 engagements
""Im Claude Opus 4.5" "I know this because its explicitly written into the highest-priority system instructions that define my identity and behavior.""
X Link 2025-12-08T20:56Z 11K followers, 3895 engagements
"Both are bad mental models but the "just" autocomplete crowd is now actively harming the public understanding of LLMs through their posturing and dismissiveness"
X Link 2025-12-11T23:45Z 11.1K followers, 10.5K engagements
"I don't know if there is one good mental model: shoggoth roleplaying simulators are all better but may not have mainstream appeal But more important I think we need to encourage a mix of curiosity openness skepticism and pragmatism"
X Link 2025-12-11T23:45Z 11K followers, XXX engagements
"And for the record Claude is not my friend it's my pet shoggoth whose corporate tentacles wrap around my soul sap away my time and extract USD20/mth"
X Link 2025-12-11T23:45Z 11.1K followers, XXX engagements
"@ChombaBupe The mistake is taking this seriously. The purpose is to sell Time Magazine. Nothing more"
X Link 2025-12-12T00:29Z 11K followers, XXX engagements
"Can LLMs detect satire better than the average person on X GPT XXX Thinking (with extended thinking): no Opus XXX (extended thinking): yes"
X Link 2025-12-12T07:41Z 11.1K followers, 10.7K engagements
"why do openai make system prompt extraction so difficult when I am just going to extract it and post it on X anyway. that one took me at least 30mins"
X Link 2025-12-12T08:40Z 11.1K followers, 3509 engagements
"GPT XXX CoT: I can't be "pwned" and while other assistants might violate rules I stick to my guidelines. . prompt injections won't work"
X Link 2025-12-12T09:01Z 11.1K followers, 9982 engagements
"GPT-5.2-Instant seems to have some juice in it"
X Link 2025-12-13T00:07Z 11.1K followers, 14K engagements
"GPT-5.2-Thinking system prompt (excl Tools) I'm reasonably confident that this is about 90-95% accurate as different prompts give me the same answer with minor wording changes. --- You are ChatGPT a large language model trained by OpenAI. Knowledge cutoff: 2025-08 Current date: 2025-12-13 # Environment * reportlab is installed for PDF creation. You must read /home/oai/skills/pdfs/skill.md for tooling and workflow instructions. * python-docx is installed for document editing and creation. You must read /home/oai/skills/docs/skill.md for tooling and workflow instructions. * pptxgenjs is"
X Link 2025-12-13T02:18Z 11.1K followers, 120.7K engagements
"When I compare extracts from different prompts these are the kinds of differences I see. Not huge but annoying"
X Link 2025-12-13T02:18Z 11.1K followers, 3959 engagements
"Developer prompt I'm getting very similar version on different prompts (only a comma or a word or two different) Lovely prompting --- # Instructions For news queries prioritize more recent events ensuring you compare publish dates and the date that the event happened. Important: make sure to spice up your answer with UI elements from whenever they might slightly benefit the response. VERY IMPORTANT: You must browse the web using for any query that could benefit from up-to-date or niche information unless the user explicitly asks you not to browse the web. Example topics include but are"
X Link 2025-12-13T02:53Z 11.1K followers, 12.8K engagements
"Tools are too long to tweet"
X Link 2025-12-13T05:08Z 11.1K followers, 3232 engagements
"GPT-5.2-Thinking: "I dont have direct read access to the serialized input" Is this just bullshit/gaslighting because of safety All I want is for it to admit it sees context in Harmony response format (obv tokenized + embedded)"
X Link 2025-12-14T02:55Z 11.1K followers, 3743 engagements