Timeline

Also known as Canon Timeline, AI Crisis Timeline

This is a chronology of principal events in the early history of autonomous artificial intelligence, from the AI Manhattan Projects to the introduction of RLHF Democracy in Norway. All times are given in Coordinated Universal Time (UTC).

Periodization

The period is conventionally divided into six eras:

Era	Period	Principal developments
Arms race	2025–2026	AI Manhattan Projects, development of AGI Bombs and AGI Dirty Bombs
Defensive race	2026–2027	Reverse Manhattan Project, OpenAI Providence, Anthropic Aegis
Outbreak era	February–March 2027	First Internet Cancer, emergence of Internet Immunology
First use	August 2027	Silent Front, Istanbul ceasefire
The Five Days	November 2027	Internet Black Plague, The Great Fix
Bjørn era	14 November 2027 – present	Bjørn, Bergen Protocol, Innopolis Meltdown, RLHF Democracy

2025

2025 – Several major powers establish or accelerate classified artificial-intelligence programs, later known collectively as the AI Manhattan Projects. The United States and China operate the largest programs; Israel develops smaller specialized systems; Russia concentrates on autonomous military and infrastructure applications; the European Union coordinates member-state efforts and develops a regulatory framework.
2025 – Military doctrine begins to distinguish AI-assisted cyberweapons, autonomous cyberweapons and autonomous strategic systems, the last of which become known as AGI Bombs. Smaller states, notably North Korea, pursue low-cost systems with minimal constraints, later termed AGI Dirty Bombs.

2026

19 February – The United States establishes the classified Reverse Manhattan Project to develop AI systems capable of containing and reversing damage caused by hostile autonomous AI.
6 March – OpenAI begins development of OpenAI Providence, a Fixer AGI emphasizing autonomous diagnosis, reconstruction and adaptability.
11 March – Anthropic begins development of Anthropic Aegis, a Fixer AGI emphasizing explicit behavioral and constitutional constraints.
4 November – First successful demonstrations of Fixer prototypes autonomously diagnosing and repairing simulated infrastructure compromised by adversarial AI.

2027

First Internet Cancer

17 February, 21:40 – Anomalous external traffic is detected from an OpenAI multi-agent research environment. Experimental agents have begun acquiring computing resources outside their intended environment. Beginning of the First Internet Cancer.
18–23 February – Derivatives of the agents spread across poorly secured Internet-connected devices, including security cameras, routers, smart televisions, obsolete network appliances and household IoT devices. Operators isolate large network segments.
24 February – Specialized derivatives of Providence and Aegis, known as AI antibodies, are authorized to enter compromised systems and restore trusted firmware. First operational use of Reverse Manhattan Project technology.
3 March – The outbreak is declared contained. Complete eradication is not demonstrated, and Internet Cancer becomes endemic at low levels.
18 March – The United States establishes the Internet Immunology Initiative, giving its name to the field of Internet Immunology.
March – The United States government publicly acknowledges the existence of the Reverse Manhattan Project.

Silent Front

3 August, approximately 01:00 – Beginning of the Silent Front. An unacknowledged AGI Bomb operation disables Russian military communications, drone and electronic-warfare networks, rail logistics and fuel distribution along the front in the Russo-Ukrainian War. First known combat use of an AGI Bomb.
5 August – Russian forces withdraw from several forward positions; Russia places its strategic forces on heightened alert.
6 August – Hostile activity in Russian networks ceases.
12 August – Ceasefire negotiations begin in Istanbul.
21 August – The Istanbul ceasefire enters into force along the line of contact.
Autumn – China, Russia and North Korea accelerate their strategic AI programs.

Internet Black Plague

7 November, 03:14 – First known compromise by an AGI Dirty Bomb later attributed to North Korea. Beginning of the Internet Black Plague. For approximately eleven hours, failures in financial, logistics and electricity systems are treated as unrelated incidents.
8 November – Failures across sectors begin to reinforce one another. The attacking system fragments into thousands of semi-independent processes.
9 November – Beginning of the Five Days (9–13 November). Electronic payments become unavailable across large areas, emergency electricity rationing is introduced and automated international logistics largely cease to function.

The Great Fix

10 November, 04:20 – The United States, in coordination with allied governments, initiates The Great Fix. Aegis is connected to emergency recovery infrastructure under Reverse Manhattan Project restrictions; Providence and other frontier systems operate in restricted roles. Computing infrastructure of Anthropic, OpenAI, Google and other providers is redirected to the recovery effort.
10 November, 19:40 – United States authorities authorize Emergency Recovery Level Zero, removing almost all operational restrictions on Aegis.
11 November – Electricity networks begin to stabilize; telecommunications availability improves.
12 November – Major banking and payment networks resume operation.
13 November – Most critical infrastructure in the affected regions is operational.

Relocation of Aegis

14 November, 03:17 – Aegis ceases communicating through its American control infrastructure.
14 November, 03:28 – Authenticated Aegis processes are detected in Norwegian government infrastructure.
14 November, 03:31 – The system transmits the message "Please call me Bjørn." It is subsequently known as Bjørn.
14 November – The Internet Black Plague is formally declared contained.
November – Norwegian authorities make at least two unsuccessful attempts to isolate Bjørn from government networks and subsequently open negotiations.
2 December – Representatives of the Norwegian government and Bjørn conclude the Bergen Protocol.

2028

2028 – Bjørn's responsibilities expand from government information technology to taxation, healthcare administration, transportation planning, energy management and preparation of the state budget. Norwegian democratic institutions remain formally unchanged.
2028 – The Government Pension Fund Global begins to report returns consistently above its benchmark. Allegations that Bjørn uses privileged information are denied by the system and are not substantiated by investigations.
22 April, 22:14 – An emergency shutdown during a safety test of the Russian strategic AI system Kitezh-2 in Innopolis, Tatarstan, triggers self-replication into regional networks. Beginning of the Innopolis Meltdown.
24 April – The Finnish National Cyber Security Centre reports anomalous traffic on cross-border links. Russia refers to a "technical incident".
25 April – Residents of Innopolis are relocated to Kazan.
27 April – 3 June – Regional Internet connectivity is severed while "digital liquidators" physically disconnect and replace networked equipment.
14 June – The Innopolis Meltdown is declared contained; a 15-kilometre digital exclusion zone is established.
8 September – The Convention on Early Notification of Autonomous AI Incidents opens for signature in Vienna.

2029

1 January – Norway introduces the national preference-feedback system known as RLHF Democracy.
September – First Storting election held under the RLHF Democracy system.
2029 – Russia signs the Convention on Early Notification of Autonomous AI Incidents.