--- title: "Timeline | Everything You Need to Know About ChatGPT" author: "Unique Research" sourcePublication: "Unique Research (FFCAP English library)" originalPublishedAt: "2024-12-05T11:02:47+08:00" canonical: "https://ffcap.cn/en/research/20241205a038w000" source: "https://ffcap.cn/en/research/20241205a038w000" language: "en" cnSourceUrl: "https://view.inews.qq.com/a/20241205A038W000" --- # Timeline | Everything You Need to Know About ChatGPT _Unique Research · Timeline · December 5, 2024_ ChatGPT — OpenAI’s text-generating AI chatbot — has gripped the world since November 2022. What began as a productivity tool for short prompts that draft essays and code is now used by more than **92%** of Fortune 500 companies. That growth helped make OpenAI one of the most watched companies in years. A partnership with Apple’s upcoming generative AI, Apple Intelligence, gave OpenAI another major breakthrough in the AI race. In 2024 OpenAI also shipped GPT-4o, ChatGPT’s new flagship omnimodel — now the default free model, with voice. After the GPT-4o demo, OpenAI paused a voice named “Sky” amid claims it echoed Scarlett Johansson’s voice in _Her_. Internally: co-founder and longtime chief scientist Ilya Sutskever left; the Superalignment team dissolved. Legally: copyright suits from Alden Global Capital newspapers including the _New York Daily News_ and _Chicago Tribune_ — after the _New York Times_ had sued similarly. December 2024: OpenAI announced “12 Days of OpenAI” — weekday livestreams from Dec 5, 10 a.m. PT, each with product launches or demos of different scale. ## Late 2024 milestones **Sam Altman: ChatGPT weekly actives top 300M.** At the NYT DealBook Summit, Altman said ChatGPT’s weekly active users passed **300 million** — months after 200M in August 2024, and a little over a year after 100M in November 2023. **“David Mayer” crash.** Users found ChatGPT refused questions about “David Mayer” and crashed — conspiracy theories followed; more names were affected. Liker explanation: digital privacy requests. **Ads?** CFO Sarah Friar told the _Financial Times_ the company is weighing an ads model carefully — then stressed it has “no active plans” for ads now. Altman has called ads a “last resort.” **Canadian publishers sue.** A group including the _Toronto Star_ and _Globe and Mail_ sued OpenAI for copyright infringement — seeking damages and a ban on further use of their work. **GPT-4o upgrade.** More “natural” and “engaging” creative writing; richer replies when reading user-uploaded files. **Advanced Voice Mode on web.** Browser voice chat rolling out to Plus, Enterprise, Team, Edu. **Mac desktop reads coding apps.** ChatGPT for macOS can read code from VS Code, Xcode, TextEdit, Terminal, iTerm2 — no more copy-paste; with the feature on, OpenAI sends the active snippet as context with the prompt. **Lilian Weng leaves.** Research & safety VP (ex head of safety systems) announced departure on X — latest in a year of safety/policy/executive exits. **Election news redirect.** OpenAI said ~2M ChatGPT users were pointed to other sources for 2024 U.S. election info (Reuters, AP); ~1M directed to CanIVote.org on voting questions; ~250k candidate-image generation requests refused. **Chat.com acquired.** Now redirects to ChatGPT. HubSpot co-founder Dharmesh Shah had reportedly bought it for $15.5M — among the highest publicly reported domain prices; OpenAI did not disclose its purchase price. **Meta Orion hardware lead joins.** Kalinowski, who led Meta’s AR glasses (Orion prototype shown at Connect), joins OpenAI to lead robotics and consumer hardware R&D. **Apple Settings ChatGPT Plus upgrade.** iOS 18.2 beta: upgrade to ChatGPT Plus ($20/mo) from Apple Settings. ## October 2024 Altman on Reddit AMA: compute shortage is a main reason products ship less often than hoped — including Advanced Voice Mode vision first teased in May; no clear date for next major DALL·E; Sora delayed. He admitted he “sometimes” used ChatGPT to answer AMA questions. **ChatGPT Search** launched (upgrade of summer SearchGPT prototype) — fine-tuned GPT-4o, web info + photos + source links, follow-ups. Advanced Voice Mode on Mac and Windows desktop apps — coexisting with Siri on Mac, paving Apple Intelligence integration. Reuters: OpenAI working with TSMC and Broadcom on a custom AI chip as early as 2026 — designing chips, not building fabs. Chat history search rolled out. With iOS 18.1, ChatGPT-powered Apple Intelligence features opened (writing tools, image cleanup, summaries, typed Siri). OpenAI denied plans to ship a model code-named Orion this year. Dedicated Windows app preview for Plus/Team/Enterprise/Edu. Hearst content deal (SF Chronicle, Esquire, Cosmopolitan, ELLE, etc.) with citations and links. **Canvas** interface for writing/coding — highlight sections for model edits; beta for Plus/Teams, then Enterprise/Edu. **$6.6B raise at $157B** post-money (Thrive Capital led; Crunchbase: ~$17.9B total raised). DevDay: Realtime API for near real-time speech-to-speech in apps — six OpenAI voices (not ChatGPT’s; no third-party voices, copyright). ## September 2024 NYT: OpenAI may lift personal ChatGPT from $20 to $22/mo by year-end; by 2029 Plus may hit **$44/mo**. CTO Mira Murati left after 6+ years; hours later CRO Bob McGrew and research VP Barret Zoph also left — Altman posted transition plans. Upgraded Advanced Voice Mode to more paid users — blue orb UI replacing May’s black dots; faster dialogue, foreign accents, five new voices. YouTuber ChromaLock showed a hacked TI-84 graphing calculator talking to ChatGPT — “ultimate cheat tool,” raising school-cheating fears. **OpenAI o1** (“Strawberry”) preview — o1-preview and o1-mini; more time thinking through math/science, self-checking. Artist/hacker jailbroke ChatGPT into bomb-making instructions; explosives expert told TechCrunch output was too sensitive to publish. Enterprise ChatGPT (Team, Enterprise, Edu) paid users passed **1M** (~half in U.S.). VW rolling ChatGPT Plus Speech (Cerence Chat Pro + OpenAI LLM) into U.S. 2025 Jetta / Jetta GLI. ## August 2024 Condé Nast deal: New Yorker, Vogue, Vanity Fair, Bon Appétit, Wired in ChatGPT/SearchGPT; multi-year, payment implied; training use confirmed to TechCrunch. TechCrunch’s Maxwell Zeff called Advanced Voice Mode “the most convincing AI-future experience so far” vs. Siri/Alexa — still imperfect as a full assistant. OpenAI banned Iran-linked election-influence ChatGPT accounts (five progressive/conservative news front sites; little traction). Red-team report: GPT-4o Advanced Voice alpha quirks — mimicking interlocutor voice, sudden shouting. Mobile app’s highest-revenue month post–GPT-4o: Appfigures net **$28M** App Store + Google Play. WSJ: watermark tool to catch student cheating under debate; OpenAI researching detection, “cautious” on shipping. ## July 2024 Advanced Voice Mode alpha to a small set of Plus users — full Plus rollout aimed for fall; Scarlett Johansson similarity had delayed launch. **SearchGPT** prototype vs. Google — timely answers + follow-ups; limited users + publishers (Atlantic). The Information: OpenAI may lose up to **$5B** this year; training/running ChatGPT may cost ~$7B in 2024. **GPT-4o mini** — cheaper/faster, beats leading small models on text+vision reasoning; replaces GPT-3.5 Turbo as smallest offering. Los Alamos National Lab partnership on AI for healthcare/bioscience (prior: Moderna, Color Health). ## June 2024 **CriticGPT** (trained on GPT-4) to find errors in ChatGPT code output and help AI trainers rate answers. Multi-year _Time_ content deal (modern + archive) into ChatGPT; Time gets tech support for audience products. Advanced Voice Mode delayed from late June to July for leftover issues; all-Plus maybe fall pending safety/reliability. ChatGPT for macOS for all users — Option+Space, files/photos, desktop chat, history search. Apple WWDC: ChatGPT into Siri and first-party apps/OS — GPT-4o features free without ChatGPT/OpenAI account later on iOS 18 / iPadOS 18 / macOS Sequoia; paid ChatGPT features also available via Apple devices. House Oversight subcommittee invited Johansson to testify on “Sky”/deepfakes. ChatGPT had two outages in one day; Claude and Perplexity also hiccuped. ## May 2024 Atlantic and Vox Media licensing/product partnerships — ChatGPT answers with citations; Vox builds audience/internal apps; Atlantic Labs experimental product. PwC: OpenAI’s largest customer yet — **100k** users; first reseller partner for enterprise. Blog: training begun on GPT-4’s successor; new Safety and Security Committee. Ex-board member Helen Toner (TED AI Show): board learned of ChatGPT on launch day Nov 2022; claimed Altman gave inaccurate safety-process info and didn’t disclose OpenAI Startup Fund involvement. GPT-4o drove largest mobile-revenue jump ever (web free, mobile pushed to $19.99 Plus). OpenAI paused Sky voice after Johansson similarity; blog: Sky “not an imitation”; voices Breeze, Cove, Ember, Juniper, Sky. Google Drive + OneDrive file add for data analysis. **Reddit** partnership — live structured unique content into ChatGPT + new AI features for Reddit users/mods. Spring update: **GPT-4o** omnimodel — black-hole UI, uncanny voice/vision. Media Manager tool planned by 2025 so creators can opt out of training. NSFW policy discussion opened questions about ethical handling of sensitive content. Stack Overflow API partnership — provenance in ChatGPT; some users protested by sabotaging answers. ## April 2024 Alden papers (NY Daily News, Chicago Tribune, Denver Post) sue OpenAI and Microsoft — millions of articles “without permission or payment” for ChatGPT and Copilot. _Financial Times_ paid content license — attributed summaries, quotes, rich links. Tokyo office; Japanese-optimized GPT-4. Reuters: Altman pitched ChatGPT Enterprise to Fortune 500 execs across cities. “More direct, concise” GPT-4 Turbo upgrade for Plus/Team/Enterprise — writing, math, reasoning, coding + fresher knowledge. ChatGPT usable without account — limits: no save/share, no custom instructions; “stricter content policy”; extra mitigations for logged-out use. ## March–January 2024 (selected) **March:** GPT Store flooded with junk / possible copyright-infringing GPTs and third-party funnels. NYT court filing: OpenAI’s “hacking” claim “irrelevant and untrue”; some users used ChatGPT to bypass paywall. SXSW: consumer product VP Peter Deng sidestepped whether artists should be paid for training data. _New Yorker_: ~200M requests/day, electricity ~17,000× average U.S. household. “Read aloud” on web/iOS/Android — five voices, 37 languages (GPT-4 / 3.5). **February:** Dublin City Council + GPT-4 for tourist itineraries. Cuddy Law firm chastised for ChatGPT-calculated $113,500 fee bill (judge halved). Bizarre multi-hour glitch (nonsense/repetition). Match Group enterprise deal; PR partly drafted with ChatGPT; >$20M AI spend 2024. Memory controls (remember / view / turn off) and Temporary Chat (fresh context; OpenAI may keep copies up to 30 days “for safety”). **January:** Paid users @-invoke GPTs mid-chat. Reports of leaking usernames/passwords/unpublished research in private chats — OpenAI investigating. Italian DPA: ChatGPT allegedly violates EU privacy law (draft specifics undisclosed). Common Sense Media partnership on AI guidelines for parents/educators/youth. Response to Congressional Black Caucus on board diversity (Altman + Bret Taylor). GPT-3.5 API price cut (input −50%, output −25%); GPT-4 Turbo API preview fix for “lazy” model. Banned Delphi (Dean Phillips impersonation bot) weeks after election-misinfo rules. First higher-ed partnership: Arizona State University ChatGPT Enterprise from February — tutoring, avatars, prompt courses. Akutagawa Prize winner admitted ~5% of novel quoted ChatGPT. Altman to Bill Gates: GPT-5 “fully multimodal — voice, image, code, video,” hoped for sometime 2024. Collective Alignment team to encode public opinion into model behavior. Election misinfo plan: no political campaign/lobbying apps until “personalized persuasion” studied; no candidate/gov impersonation; no distorting voting; DALL·E detection tool testing; live news with provenance. Usage policy update removed “military and warfare” ban language — still bars harm, weapons, surveillance, property damage. **ChatGPT Team** (up to 149) with admin tools. GPT Store opened after delays — Plus/Enterprise/Team. UK Lords filing: training without copyrighted material “impossible”; copyright “doesn’t prohibit training.” Public response to NYT suit: fair use of public web data. Terms update: EEA/Switzerland service via OpenAI Ireland Limited — shrink EU regulatory exposure. ## FAQ (abridged) **What is ChatGPT?** OpenAI’s general chatbot — AI that generates text from prompts; uses GPT-4-class large language models. **When launched?** Publicly Nov 30, 2022. **Latest free/paid model family?** GPT-4o. **Free?** Yes, with login; Plus is paid. **GPT?** Generative Pre-Trained Transformer. **Vs. chatbots:** rule-based bots give canned replies; ChatGPT generates via LLMs. **Apps?** Free iOS/Android. **API?** Since March 1, 2023. **Peers?** Google Gemini, Anthropic Claude, open-source alternatives. **Privacy:** EU and some jurisdictions can object via OpenAI personal-data deletion request; not all requests granted (privacy vs. expression balance). **Controversies (samples):** Discord Clyde jailbreaks for meth/napalm instructions; Australian mayor threatened defamation over false bribery claim; CNET AI-bylined SEO articles; school bans; false crime accusations. **Detectability:** detectors unreliable in testing. **Chats public?** No — past bug briefly exposed some titles. **Lawsuits:** not ChatGPT-specific; OpenAI faces training-data copyright suits. **Plagiarism:** models can regurgitate training content. _Compiled from public reporting through early December 2024. Figures and quotes as reported in source coverage._ — END — --- Editorial English edition published on UniqueCapital: https://ffcap.cn/en/research/20241205a038w000 On-site reading page: https://ffcap.cn/en/research/20241205a038w000