Before griefbots: early chatbots and lifelogging
1966
January 1966: Joseph Weizenbaum of MIT describes ELIZA in the journal Communications of the ACM. ELIZA matched keywords to scripted replies, most famously imitating a psychotherapist, and Weizenbaum was troubled by how easily people believed it understood them.
2001
2001: Microsoft Research begins MyLifeBits, a lifelogging experiment in which the computer scientist Gordon Bell digitises his documents, photos, recordings and messages. The project showed that a whole life could be stored and searched, long before AI could talk about it.
Fiction and first plans (2013 to 2015)
2013
11 February 2013: Channel 4 broadcasts the Black Mirror episode “Be Right Back”, in which a grieving woman talks to an AI built from her late partner’s online messages. It became the cultural reference point for almost every later griefbot. See what in “Be Right Back” is real now.
2014
2014: Entrepreneur Marius Ursache and a team from an MIT entrepreneurship programme announce Eterni.me, which promised an avatar built from a person’s digital traces. The Verge reported in 2016 that it had not launched as a product.
2015
November 2015: Google releases its TensorFlow machine-learning library as open source. Less than three weeks later, on 28 November, Roman Mazurenko is killed by a car in Moscow.
Memorial bots built by hand (2016 to 2019)
2016
24 May 2016: Eugenia Kuyda announces a memorial chatbot of her friend Roman Mazurenko, trained on more than 8,000 lines of messages from ten friends and relatives. It is one of the best-documented early griefbots. Read the full Roman Mazurenko story.
2016: Journalist James Vlahos records his dying father’s life story and scripts it into a “Dadbot” on Facebook Messenger. See James Vlahos’s Dadbot.
2016: According to NPR, Ray Kurzweil begins building a chatbot from his late father’s letters and lectures, later called the Fredbot. See Ray Kurzweil’s father bot.
2017
11 April 2017: Microsoft files a US patent application for creating a chatbot of a specific person from their social data. It is published in October 2018.
July 2017: WIRED publishes Vlahos’s account of the Dadbot. The same year, Kuyda’s company releases Replika, a companion app that grew out of the Roman bot experience.
2019
March 2019: Vlahos publishes Talk to Me, a book on voice computing that includes the Dadbot story. He later co-founds HereAfter AI, which lets families hear recorded answers from a loved one.
The language model era (2020 to 2022)
2020
February 2020: Korean broadcaster MBC airs the documentary “Meeting You”, in which a mother, Jang Ji-sung, meets a virtual reality recreation of her seven-year-old daughter, Nayeon, who died in 2016. Read our page on Meeting You.
Summer 2020: Game designer Jason Rohrer launches Project December, a chat site powered by OpenAI’s GPT-3. On 24 September 2020 Joshua Barbeau uses it to simulate his fiancée Jessica Pereira, who died in 2012. See the Project December story.
1 December 2020: Microsoft’s patent US10853717B2 is granted. Read what the Microsoft patent describes.
2021
January 2021: After the patent makes headlines, a Microsoft AI manager says there is no plan to build it and calls the idea disturbing.
23 July 2021: The San Francisco Chronicle publishes “The Jessica Simulation”. In September 2021 OpenAI ends Project December’s access after Rohrer refuses to add the monitoring and restrictions it requested.
2022
22 June 2022: Amazon demonstrates Alexa reading a story in a child’s late grandmother’s voice, learned from less than a minute of audio. The demonstration sharpened debate about voice cloning and consent.
October 2022: MIT Technology Review reports on recorded-story services such as HereAfter AI and StoryFile. In the same month The Register reports that Project December has relaunched as a service for simulating people who have died.
Research, rules and a growing market (2023 to 2026)
2023
2023: A study presented at the CHI conference interviews ten mourners who used chatbots, finding both comfort and risks. Amy Kurzweil publishes Artificial: A Love Story, a graphic memoir about the Fredbot.
2024
20 January 2024: The documentary Eternal You, by Hans Block and Moritz Riesewieck, premieres at the Sundance Film Festival. See our page on Eternal You.
May 2024: MIT Technology Review reports that at least half a dozen companies in China offer AI avatars of dead relatives, and that thousands of people have paid for them. In the same month, Cambridge researchers publish a paper on “deadbots” calling for consent, age limits, transparency and dignified ways to retire them.
1 August 2024: The EU Artificial Intelligence Act enters into force. It requires, among other things, that people are told when they are interacting with an AI system.
2026
Mid-2026: HereAfter AI announces that it is shutting down and asks users to request their recordings. See our HereAfter AI comparison.
2 August 2026: The AI Act’s transparency duties for chatbots begin to apply in the European Union. Laws differ by country; this is general information, not legal advice.
What the timeline shows
| Period | How bots were made | Main concern |
|---|---|---|
| 1966 to 2015 | Scripts, archives and fiction | People believing a program understands them |
| 2016 to 2019 | Hand-built bots from messages, interviews or letters | Consent, curation and how it feels for family |
| 2020 to 2022 | Large language models and voice cloning | Fluent invention, dependence and loss of access |
| 2023 to 2026 | Commercial services and regulation | Business models, transparency and safe endings |
Two things stand out. Each step made building a simulation easier, from a team of engineers in 2016 to a single person in an evening in 2020. And the questions raised by the earliest projects, about consent, honesty and how to stop, are the same ones researchers and regulators are still working on.
Where Memory Clone fits
Memory Clone is a recent, modest entry in this story. It is a Windows app, in alpha, that builds a private AI simulation from chats you export yourself. Originals stay on your PC, check-ins are off by default, and every clone is presented as an AI simulation, not the person. It does not use voice, photos or video. If you want to read more, visit the download page. There is no need to decide anything now.
If you are in crisis or thinking about harming yourself, contact your local emergency number now. You can also reach: UK & Ireland Samaritans, 116 123; United States 988 Suicide & Crisis Lifeline, call or text 988; elsewhere find a free, confidential helpline at findahelpline.com.
Frequently asked questions
What was the first griefbot?
One of the best-documented early examples is the Roman Mazurenko bot, announced in May 2016. Earlier ideas existed, such as Eterni.me in 2014, and fiction imagined it in 2013.
Was ELIZA a griefbot?
No. ELIZA was a general scripted chatbot from 1966. It matters to this history because it showed how readily people treat a chatbot as if it understands them.
When did griefbots become common?
From 2020, when large language models such as GPT-3 let anyone create a fluent imitation from a few messages. By 2024, reporting described commercial griefbot services in several countries.
Are griefbots regulated?
There is no single griefbot law. In the EU, the AI Act requires chatbots to disclose that they are AI, with these duties applying from August 2026. Laws differ by country; this is general information, not legal advice.
Sources
Checked on . Third-party products and help pages change often; follow the links for current details.
- Weizenbaum, “ELIZA: a computer program for the study of natural language communication between man and machine” (Communications of the ACM, 1966) (ACM)
- MyLifeBits (Microsoft Research)
- Speak, Memory (Casey Newton, 2016) (The Verge)
- A Son’s Race to Give His Dying Father Artificial Immortality (James Vlahos, 2017) (WIRED)
- Mother Meets Recreation of Her Deceased Child in VR (Scott Hayden, 2020) (Road to VR)
- US10853717B2: Creating a conversational chat bot of a specific person (Google Patents)
- The Jessica Simulation: Love and loss in the age of A.I. (Jason Fagone, 2021) (San Francisco Chronicle)
- Amazon demonstrates Alexa mimicking the voice of a deceased relative (2022) (CNBC)
- Deepfakes of your dead loved ones are a booming Chinese business (Zeyi Yang, 2024) (MIT Technology Review)
- Hollanek & Nowaczyk-Basińska, “Griefbots, Deadbots, Postmortem Avatars” (Philosophy & Technology, 2024) (Springer)
- Regulation (EU) 2024/1689 (Artificial Intelligence Act) (EUR-Lex)
This article is general information, not medical, legal or financial advice. Memory Clone creates an AI simulation from the material you provide; it is not the person.