[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.]
@koltregaskes Kol Tregaskes
Kol Tregaskes posts on X about open ai, 6969, gpt, microsoft the most. They currently have XXXXXX followers and XXX posts still getting attention that total XXXXXX engagements in the last XX hours.
Engagements: XXXXXX #
- X Week XXXXXXX +14%
- X Month XXXXXXXXX +145%
- X Months XXXXXXXXX +34%
- X Year XXXXXXXXX +656%
Mentions: XX #
- X Week XXX +13%
- X Month XXX +53%
- X Months XXXXX -XX%
- X Year XXXXX +658%
Followers: XXXXXX #
- X Week XXXXXX +2.10%
- X Month XXXXXX +4.20%
- X Months XXXXXX +48%
- X Year XXXXXX +228%
CreatorRank: XXXXXXX #
Social Influence #
Social category influence technology brands XXXXX% stocks XXXX% finance XXXX% travel destinations XXXX% countries XXXX% celebrities XXXX%
Social topic influence open ai #434, 6969 3.86%, gpt #38, microsoft #140, agi 1.93%, o3 1.93%, faster 1.45%, $googl 1.45%, xai 1.45%, has been XXXX%
Top accounts mentioned or mentioned by @idontexist_nn @bughunternerd @openai @idontexistnn @lalopenguin @grok @chatgpt21 @cengizdemiurg @testingcatalog @katherinebrain @tombombadeel @aimelgibson @1043 @shawnryan96 @sama @btibor91 @kimmonismus @midjourney @sadalbert10 @wltnwang
Top assets mentioned Microsoft Corp. (MSFT) Alphabet Inc Class A (GOOGL) IBM (IBM) Dave Inc. Class A Common Stock (DAVE) Salesforce Inc (CRM)
Top Social Posts #
Top posts by engagements in the last XX hours
"@idontexist_nn @bughunternerd @____1043 @lalopenguin Yes I've mentioned this before once we get there of course "AGI-1" should be the same. Boring but logical"
@koltregaskes on X 2025-07-23 22:57:53 UTC 12.8K followers, XX engagements
"@idontexist_nn @bughunternerd @lalopenguin Oh I remember those names oh man nah. GPT-X could be GPT-10. GPT-Next next to what ;-)" @koltregaskes on X 2025-07-23 22:48:27 UTC 12.9K followers, XXX engagements
"Runway introduced Aleph today an in-context video model that edits input videos via text prompts to add remove or transform objects. Generate new scene angles modify styles and lighting and handle various other tasks with access rolling out gradually over the next few days" @koltregaskes on X 2025-07-25 19:42:06 UTC 12.9K followers, 1254 engagements
"8/ Collaboration & Sharing - Shared workspaces (Study Togetherstyle) where multiple users and agents collaborate live. - Realtime coediting of chats docs code with presence indicators. - Shareable dashboards / task boards with rolebased vieworedit permissions. - Oneclick export of conversations or projects as Markdown PDF or PowerPoint. - Inline comments & discussion threads attached to any message task or snippet. - Chainofthought reveal toggle for audit or teaching moments. - Publiclink view mode with expiry dates and watermarks for external stakeholders. o3' suggestions: - Live voicecall" @koltregaskes on X 2025-07-22 02:30:19 UTC 12.8K followers, XXX engagements
"My GPT-5 Wish List This is a long GPT-5 (and beyond) wish-list. I've been meaning to post this for a very long time but I think time is running out so you'll have to except it as it is right now (incomplete needing reviewing edits clearer/clarifcations). I've listed my suggestions first with o3's ideas after in each category (just for fun). Please add your ideas below: Core Models & Performance - 1Mtoken context window (on par with GPT4.1 / Gemini Ultra). - Increased accuracy and reliability better instruction-following less hallucinations - Improved emotional intelligence - First world model" @koltregaskes on X 2025-07-22 02:30:17 UTC 12.9K followers, 13.8K engagements
"@MRAJPalmer Yes Prompt Engineer has been around for a couple of years now but the need for prompting (rather than normal human communication) us slowly dying as the models get better are understanding what we type/say" @koltregaskes on X 2025-07-17 18:39:09 UTC 12.9K followers, XX engagements
"@bughunternerd @____1043 @idontexist_nn @lalopenguin Normies are not going to care about even the "GPT" thing. It needs to go away eventually. Even "ChatGPT" is a mouthful but you know marketing and stuff. ;-)" @koltregaskes on X 2025-07-23 22:53:06 UTC 12.9K followers, XXX engagements
"@D5DUBV Yes if I start a non-agent chat then it's gone as an option. Really annoying. I don't know who to tag at OpenAI about this so I'll go to my default bestie lol: @TheRealAdamG" @koltregaskes on X 2025-07-23 12:24:58 UTC 12.9K followers, XX engagements
"@sama 21/ And I asked ChatGPT to put the wish list into a report (might have another report to add here tomorrow):" @koltregaskes on X 2025-07-22 02:30:22 UTC 12.8K followers, XXX engagements
"I hope whatever AI releases are coming this year they are done before Battlefield X launches as I'll be hammering this game. Reveal trailer just popped MP reveal on 31st" @koltregaskes on X 2025-07-24 15:19:08 UTC 12.9K followers, XXX engagements
"ChatGPT Agent Plus EU rollout er rolling back Several EU reporting the same thing perhaps someone pressed the button too soon. EU users do you have Agent or not" @koltregaskes on X 2025-07-22 08:40:38 UTC 12.9K followers, 6540 engagements
"@kimmonismus No it doesn't. It just requires age verification. This is for creator monetisation" @koltregaskes on X 2025-07-26 08:46:38 UTC 12.9K followers, XXX engagements
"Mark your calendar for the 31st July open source AI glasses" @koltregaskes on X 2025-07-24 02:42:00 UTC 12.9K followers, 1747 engagements
"It will likely always be a staggered release. The servers would need a slow rollout like we have with the Agent. 😀 OAI haven't predicted capacity well up until now it must be hard to tbh. GPT-5 could be huge (based just on the hype alone and the coverage it will inevitably receive) so even a fully operational Stargate might struggle. 😜" @koltregaskes on X 2025-07-23 16:24:59 UTC 12.9K followers, XX engagements
"Claude 3-Sonnet XXX and XXX have just been removed from Anthropic API. Full names: claude-3-sonnet-20240229 claude-2.1 claude-2.0" @koltregaskes on X 2025-07-21 16:20:24 UTC 12.9K followers, 3614 engagements
"GPT-5 spotted in CoPilot. Will history repeat itself and Microsoft "launch" the new GPT first Will we get Sydney XXX Time to use up that free Copilot Pro trial methinks. 😉" @koltregaskes on X 2025-07-25 06:02:37 UTC 12.9K followers, 23.7K engagements
"@HarshithLucky3 Added to my calendar. ;-)" @koltregaskes on X 2025-07-23 20:35:24 UTC 12.9K followers, XX engagements
"6/ Advanced Voice Mode (AVM) - Lift the current message cap so AVM can run all day in the background. - Deliver faster replies; minimise deadair pauses. - Keep the audio cue during web searches to show its still thinking. - Default to local (UK) time when quoting events; no US time slips. - Handsfree camera activation for quick visual inputs midchat. - Builtin task & note tracker synced with central memory. - Adaptive verbosity: concise by default verbose only on request. - Handle selftalkers gracefully; keep listening during rambles. just for me ;-) - Separate voicestyle & personality editor" @koltregaskes on X 2025-07-22 02:30:18 UTC 12.8K followers, XXX engagements
"More connectors for non-US ChatGPT users I've seen these in the US but I'm on Pro in the UK and still don't have the extra connectors. I have just GitHub. In the US I can see: GitHub Google Drive (with sync) Gmail Google Calendar Microsoft OneDrive (personal) Box Dropbox SharePoint HubSpot Linear Outlook Calendar Outlook Email Teams Plus a Create button for custom (MCP) connections. I'm hoping this means OpenAI is trying to get feature parity across regions before the big GPT-5 release" @koltregaskes on X 2025-07-23 12:16:08 UTC 12.9K followers, 1416 engagements
"@cengizdemiurg Good luck with your launch. 😀" @koltregaskes on X 2025-07-24 15:58:18 UTC 12.8K followers, XX engagements
"OpenAI taps Oracle for another XXX GW of AI firepower. - XXX GW more capacity agreed with Oracle lifting Stargates US pipeline beyond X GW. - Sites will run X million-plus Nvidia chips once fully built. - Part of a planned $XXX B XX GW build-out over the next four years. - Abilenes Stargate I already online in parts; first GB200 racks landed last month and are training early models. - Expansion is projected to create 100000+ US jobs across construction operations and supply chains. - SoftBank CoreWeave and Microsoft stay in the mix as key compute and cloud partners. Stargates rapid scaling" @koltregaskes on X 2025-07-22 12:16:11 UTC 12.8K followers, 3701 engagements
"Kimi-K2 tops EQ-Bench the benchmark that measures emotional intelligence" @koltregaskes on X 2025-07-13 06:07:21 UTC 12.8K followers, 97.4K engagements
"Unitree introduces R1 Intelligent Companion humanoid robot priced from just $5900 under $6k Almost affordable" @koltregaskes on X 2025-07-25 10:04:44 UTC 12.9K followers, XXX engagements
"Grok X is expensive when considering the number of tokens it generates ranking second only to Claude X Opus Thinking. "Grok X needed 110m tokens o3 needed 48m and Opus X only 26m (though it ranked lower). We know that test-time scaling does achieve better results and Grok X likely learned through RL that more tokens mean better scores. But if you're an API customer you're the one paying for these better results."" @koltregaskes on X 2025-07-14 21:05:34 UTC 12.9K followers, 5958 engagements
"@bughunternerd @idontexist_nn @MaeskiPhilipi Gosh no way longer. Investors still investing. They'll want their payday. OAI are tied up with MS until 2030 too. Need to clean that up. 😉" @koltregaskes on X 2025-07-23 16:06:30 UTC 12.9K followers, XXX engagements
"550k GB200s & GB300s will start going online at xAI's Colossus X supercluster in a few weeks" @koltregaskes on X 2025-07-22 18:47:39 UTC 12.9K followers, 5222 engagements
"@techhalla Midjourney stole from artists Is this serious" @koltregaskes on X 2025-07-25 08:29:36 UTC 12.9K followers, 3033 engagements
"@Devolli7 Ha possibly. :-) Yes AVM needs a lot of work. General genAI is too restricted hopefully they can relax it soon" @koltregaskes on X 2025-07-22 22:24:57 UTC 12.9K followers, XX engagements
"@idontexist_nn @lalopenguin I think it was updated in 2024 but released in 2023 if we're talking about the same paper - the levels of AGI is the paper I point every person who raises the "AGI" word to. It's not perfect as there's way more nuance to it but it's a good simple explanation of "AGI"" @koltregaskes on X 2025-07-23 22:27:40 UTC 12.8K followers, XX engagements
"5/ Automation & Scheduling - Deep Research cron jobs with XX parallel schedules; easy calendar picker. - Agent task scheduler: queue jobs set runwindows pause/resume anytime. - Static remote machine pool for Agent; autospin instances as tasks demand. - Run tasks unattended then deliver concise digests via chat / email / push. - Conditionbased triggers (when new email with invoice parse & file). - NotebookLMstyle daily news podcast: script generate vocals playable in app - Hierarchical priorities & retry logic (autorerun on failure escalate if blocked). - Task history log with timestamps" @koltregaskes on X 2025-07-22 02:30:18 UTC 12.8K followers, XXX engagements
"It's Monday OpenAI Plus and Team users should get ChatGPT Agent from today:" @koltregaskes on X 2025-07-21 15:55:00 UTC 12.9K followers, 100.7K engagements
"OpenAI's Stargate to use XX gas turbines to generate nearly 1000MW (1GW) to power 500k GB200 cards. Elon's xAI Colossus cluster is doing a similar thing" @koltregaskes on X 2025-07-24 06:11:34 UTC 12.9K followers, XXX engagements
"Netflix uses AI to cut costs in The Eternaut TV show. Though I believe Disney Lionsgate and AMC Networks have already experimented with the same AI tool before Netflix" @koltregaskes on X 2025-07-23 08:26:00 UTC 12.9K followers, 1113 engagements
"9/ Security & Privacy - Central OAuth vault: manage and revoke credentials from a single pane. - Finegrained permission system for every agent connector and memory zone. - Endtoend encryption for chat data stored memory and exported files. o3' suggestions: - Ondevice / offline mode for sensitive queries; nothing leaves the machine. - Sandboxed code execution with automatic unit tests and dependency checks. - Peritem privacy controls: redact lock or set forget after X days. - Onetap export / delete of all stored data (right to erasure). - Enterprise compliance layer with connectorspecific logs" @koltregaskes on X 2025-07-22 02:30:19 UTC 12.8K followers, XXX engagements
"I can see this happening logical to me. You'll get: - Pool GPUs/TPUs train bigger models for less money. - Share datacentres lower power draw and CO. - One safety stack faster redteam tests and fixes. - Codesign chips cheaper silicon with safety hooks builtin. - Speak to regulators together clear rules no armsrace rush" @koltregaskes on X 2025-07-23 01:00:01 UTC 12.9K followers, XXX engagements
"Midjourney has a new official YT page link in the first comment" @koltregaskes on X 2025-07-23 20:05:25 UTC 12.9K followers, 1106 engagements
"@OpenAI I wish you had a plugin for Teams for recording meetings" @koltregaskes on X 2025-06-04 17:44:18 UTC 12.9K followers, 61K engagements
"In the next few weeks OpenAI will give us: GPT-5 GPT-5-mini GPT-5-nano o3-mini-level open weights model July and August are going to be a whale of a time. 🐋" @koltregaskes on X 2025-07-24 23:33:00 UTC 12.9K followers, 16.1K engagements
"I don't think GPT-5 will be this week. OpenAI are still rolling out Agent sorry folks maybe next week" @koltregaskes on X 2025-07-23 14:05:00 UTC 12.9K followers, 39.6K engagements
"Sam Altman felt useless relative to GPT-5 when it answered a question perfectly that he couldn't. If Sam is testing the new OpenAI model then you know it's not far from release" @koltregaskes on X 2025-07-23 21:28:05 UTC 12.9K followers, 44.8K engagements
""I am very interested in what it means to give everybody on Earth like a free copy of GPT-5 running for them all the time" - Sam Altman at the Federal Reserve Vice Chair for a fireside chat with Supervision Michelle Bowman" @koltregaskes on X 2025-07-22 19:37:04 UTC 12.8K followers, 5765 engagements
"Lots of new models in LMArena: nightride-on EB45-vision qwen3-235b-a22b-instruct-2507 nightride-on-v2 EB45-turbo folsom-07172025-1 "nightride-on-v2" claims to be from Google DeepMind" @koltregaskes on X 2025-07-22 22:41:33 UTC 12.9K followers, 2268 engagements
"@Ainthropos @idontexist_nn Smart is useful accuracy reliable can follow instructions has a huge memory and context window are more useful until they point these are "solved" then the fun begins. ;-)" @koltregaskes on X 2025-07-23 22:25:11 UTC 12.9K followers, XXX engagements
"I think Sam is confusing two things here. He's talking about hobbies but labelling them as "jobs". I'm struggling to envision many long-term jobs (excluding those involving AI management or maintenance) that will be created. We need to stop calling them new jobs and start calling them new hobbies or activities. Short-term I see new roles like: - AI Trainers: training AI in specific knowledge - AI Supervisors: managing and maintaining AI agents But once AI can run by itself I'm not seeing much for lots of people to do. What new jobs can you see in the ASI era" @koltregaskes on X 2025-07-16 23:44:00 UTC 12.8K followers, 2861 engagements
"ChatGPT Agent rolling out to Plus and EU users now it seems" @koltregaskes on X 2025-07-22 06:10:05 UTC 12.9K followers, 40.3K engagements
"To be fair everyone has their own definition and that's fine for the most part. For me we hit AGI with ChatGPT (general AI) and we now arguing about the level we're at (see DeepMind's paper on the levels of AGI). But everyone will have their own needs from the AI. I hate the term myself it's a marketing word now meaningless. I think it's straight to ASI which is a little clearer" @koltregaskes on X 2025-07-23 20:53:29 UTC 12.9K followers, XXX engagements
"Firms are signing (or not) the EU AI Code of Practice in readiness for 2nd August 2025. What the latest: ✅ OpenAI signed ✅ Mistral AI signed Anthropic intends to sign Microsoft likely to sign ❌ Meta refusing ❔ Alphabet / Google no public commitment ❔ Amazon still evaluating ❔ Apple no statement ❔ NVIDIA no statement ❔ IBM no statement" @koltregaskes on X 2025-07-21 15:32:48 UTC 12.9K followers, 2914 engagements
"It seems Civitai has announced plans to block all UK users starting July XX 2025 due to challenges in complying with the UK government's new Online Safety Act. The platform appears to have cited difficulties in meeting the strict regulations concerning content moderation age verification and risk assessments which could result in severe penalties for non-compliance. While the UK block hasn't yet taken effect Civitai's statement (at least from the screenshot) indicates this is their only viable option to avoid legal issues and maintain their operations. Note there is no official announcement" @koltregaskes on X 2025-07-20 11:30:22 UTC 12.9K followers, 1911 engagements
"ChatGPT Agent now rolling out to OpenAI's Team plans it seems. Anyone who still doesn't have it Anyone who does what are your thoughts and use cases All Pro users should have it Plus started rolling out a couple of days ago" @koltregaskes on X 2025-07-24 05:58:31 UTC 12.9K followers, 7638 engagements
"@ptrckwnsch Yeah must have been a misconfigured release" @koltregaskes on X 2025-07-23 08:35:20 UTC 12.9K followers, XXX engagements
"I expect more frequent upgrades from OpenAI after GPT-5 they are eliminating monthly 4o updates and separate GPT and o-series lines. They'll still create experimental models occasionally but the main model will have a streamlined training and development cycle. We should see the IMO model/RL techniques hopefully appear before 2026. GPT-5.5 With Stargate partially operational GB200s delivered and more compute available GPT-6 likely won't take 2-3 years" @koltregaskes on X 2025-07-25 16:47:00 UTC 12.9K followers, 6211 engagements
""GPT-5 is smarter than us in almost every way"" @koltregaskes on X 2025-07-25 20:34:00 UTC 12.9K followers, 5389 engagements
"@NZ2257 @idontexist_nn He beat me to it but yeah this is not a release just a result from a training run (which are run a lot) if successful if nothing else changes in the meantime if the GPUs turn on on time. Many caveats. ;-)" @koltregaskes on X 2025-07-23 22:45:54 UTC 12.9K followers, XX engagements
"@idontexist_nn @bughunternerd @____1043 @lalopenguin But AGI-Dave-1. Has to have Dave in there somewhere else I won't use it" @koltregaskes on X 2025-07-23 22:58:34 UTC 12.8K followers, XX engagements
"The UK government's Online Safety Act adds more protection today this today. Companies now legally need to protect children from harmful material online. This has resulted in some websites closing to UK users including Civitai which now shows this on their website:" @koltregaskes on X 2025-07-25 11:05:00 UTC 12.9K followers, 2600 engagements
"Personality creator or selector coming to ChatGPT I hope both; selector for normies and something advanced for the rest of us" @koltregaskes on X 2025-07-22 15:23:00 UTC 12.8K followers, 4559 engagements
"Mark Zuckerberg says we are seeing the first glimpses of self-improvement and so the development of superintelligence is now in sight" @koltregaskes on X 2025-07-20 20:43:00 UTC 12.9K followers, 43.1K engagements
"UK signs voluntary MoU with OpenAI to pilot ChatGPT in public services and build homegrown AI infrastructure. On 21st July 2025 Technology Secretary Peter Kyle and OpenAI CEO Sam Altman inked a nonbinding memorandum to explore deploying advanced models across justice defence education and the wider civil service promising faster admin and smarter GOV.UK guidance. The partnership also investigates X bn AI Growth Zones potential UK datacentre investments and a deeper safetydata exchange with the AI Security Institute while OpenAI plans to grow its London engineering hub and run skillbuilding" @koltregaskes on X 2025-07-22 23:28:00 UTC 12.8K followers, 1112 engagements
"Something coming from Devin Place your bets what it is:" @koltregaskes on X 2025-07-22 13:03:00 UTC 12.8K followers, 1077 engagements
"@LuCaPloo Very much I chuckle when people said o3 was AGI in December. Ah man" @koltregaskes on X 2025-07-23 20:48:03 UTC 12.9K followers, XXX engagements
"Qwen3 235B Thinking or qwen3-235b-a22b-thinking-2507 launch imminent" @koltregaskes on X 2025-07-25 08:40:42 UTC 12.9K followers, 1277 engagements
"@jordandchesney @midjourney From previous experience this will improve but they'll likely not go fully uncensored. It's definitely set too high right now though I've had some weird blocks. Give them feedback so it learns. 😀" @koltregaskes on X 2025-07-24 06:02:37 UTC 12.9K followers, XXX engagements
"GPT-5 won't be AGI but everyone on here will claim it is. The clickbaiting will be wild. I can see it now. 😜" @koltregaskes on X 2025-07-23 20:30:00 UTC 12.9K followers, 29.4K engagements
"4/ Memory & Personalisation - Unified + larger longterm memory vault: fully searchable and wipeable on demand. - Automatic capture: read my email memorise everything relevant track notes in real time. - Proactive recall: surface context or reminders without my prompting when it helps the current task. - Custom Instructions 2.0: separate global projectlevel and chatlevel personas/settings. - Full personality customisation: build bespoke voices (voice editor / ElevenLabs) and character quirks independent of system instructions. - Mood slider: tweak formality humour verbosity live without" @koltregaskes on X 2025-07-22 02:30:18 UTC 12.8K followers, XXX engagements
"@jordandchesney @midjourney Tweaking the words can help" @koltregaskes on X 2025-07-24 06:03:19 UTC 12.9K followers, XXX engagements
"We have another mysterious video model called "Waver 1.0" from ByteDance dropped in 3rd on Video Arena" @koltregaskes on X 2025-07-21 12:33:31 UTC 12.9K followers, 4556 engagements
"12/ Pricing & Billing - Mid-tier plan $50-70/month - Buy credits for month tasks o3' suggestions: - XX % cheaper API pricing for GPT5 and future models. good luck with that - Clear realtime token / cost meter visible in chat and task dashboards. - Granular tiering (Free / Plus / Pro / Teams / Enterprise) with feature toggles per tier. - Payperagent compute: only pay heavier rates when spinning up large subagents or remote Operator machines. - Usage caps spend alerts and autopause when a project hits its budget. - Monthly cost digest email summarising token use tool invocations and storage. -" @koltregaskes on X 2025-07-22 02:30:20 UTC 12.8K followers, XXX engagements
"Reinforcement beats brute data in the AI arms race. - LLM gains are stalling so labs now reward-and-punish models (RLHF) instead of just stuffing them with more text. - Classic RLHF shines in single-answer domains like maths; open-ended tasks get richer mid-task feedback (think building a 3D Tetris clone). - Researchers are spinning up digital gyms - sandbox clones of apps such as Salesforce DoorDash and Amazon - where models learn to click type and buy on their own but every action burns pricey GPU time. - RL has been a post-script to pre-training yet xAI and others now channel equal compute" @koltregaskes on X 2025-07-26 11:55:00 UTC 12.9K followers, XXX engagements
"10/ Developer & PowerUser Tools - RepoPrompt workflow: ingest an entire coderepo edit & refactor multiple files in one shot. - UI designer interface like Lovable etc. - Oneclick fix bugs button: run static analysis security checks and autopatch suggestions. Codex kinda does this - Agent dev sandbox: record UI actions export as reproducible scripts for other agents. o3' suggestions: - Version pinning & diff: lock a specific model/toolrevision and compare outputs across versions. - Promptengineer toolkit: syntax highlighter similarity search prompthistory diff viewer. - API trace & cost" @koltregaskes on X 2025-07-22 02:30:19 UTC 12.8K followers, XXX engagements
"EU ChatGPT Pro users all now have Agent. Plus users rollout has started will take a few days. Have you got ChatGPT Agent now" @koltregaskes on X 2025-07-23 06:38:33 UTC 12.9K followers, 25.3K engagements
"@Skillzx5_ And also that it's either or. I work I'm using both existing automation and AI where I can which is currently very little and mostly helping me with PowerQuerys and VBA code" @koltregaskes on X 2025-07-23 20:58:02 UTC 12.9K followers, XXX engagements
"@chatgpt21 As in we get stuff every day I mean Chris" @koltregaskes on X 2025-07-23 16:20:38 UTC 12.9K followers, XXX engagements
"Persuasion tricks more than double GPT-4o-minis willingness to break its own rules. Sweet-talking GPT-4o-mini boosted its rule-breaking compliance from XXXX% to XXXX% across 28000 prompts when laced with classic persuasion cues - authority commitment scarcity social proof and more - exposing parahuman suggestibility and fresh safety headaches" @koltregaskes on X 2025-07-25 12:06:00 UTC 12.9K followers, XXX engagements
"I had this post sitting here for a day and a half well before the Verge story dang it. 😆" @koltregaskes on X 2025-07-25 02:50:00 UTC 12.9K followers, 1716 engagements