AI Accords and the NEOnix Incident

The AI Accords were a series of laws passed due to the NEOnix Incident in 2440. The Accords made it illegal for UN or UEG military forces to construct AI soldiers or other constructs without going through several feet of red tape, effectively neutering the UNSC's attempts to develop drone soldiers to aid regular ones. This would later be circumvented by private military companies such as Blacklist Private Military Contractors which the makers of the Accords had not considered restricting like they had the UNSC.

NEOnix Incident
In 2438, the UN showed off its first robotic soldier, the NEOnix Auxiliary T1C5, or the NAT to the general public. Many were astonished to see how human like the robot was and predictions on the future of robot warfare were soon running abound. The only model the UNSC had was lovingly dubbed Gypsy and had grown to become friends with its creators, being built on a smart AI in order to make it even more effective than the standard soldier.

The future looked hopeful but one mistake soon brought NEOnix's Auxiliary Project to its knees. The NEOnix company in Sarajevo had been developing the model for seven years for the UNSC's use but due to budget restrictions, was only able to secure itself one smart AI for its model. Without a proper way to get a new one, and not wishing to present a sub par model by removing the AI and thus decided to attempt to stretch out its lifespan as much as possible. The Riemann Matrix was modified in order to not automatically shut the AI down after its set lifespan and Gypsy lived far longer than he should have.

This came crashing down when in 2440, nine years into Gypsy's life, went rampant, going insane in the middle of a test for several UNSC officials. Gypsy broke out of the firing range it had been firing in, killed two nearby guards, and proceeded to go on a rampage through the NEOnix labs for two hours. Over the course of the massacre, 40 scientists were murdered along with 53 guards, 7 high ranking UNSC officials, 13 of their bodyguard and 21 of the officers who had attempted to stop the AI.

Gypsy was finally overwhelmed and destroyed by local law enforcement, the entire city of Sarajevo having gone into lockdown due to the event. The aftermath of the killings lead to to the AI Accords, as well as hundreds of studies on the nature of rampancy and the use of AI.

AI Accords
In 2440, after several months of gathering evidence and uncovering the story of the Auxiliary Project, NEOnix was put on trial along with several UNSC officials who had turned a blind eye to the companies practices. The trials lasted for three months, the trials evolving from a decision on the punishment for the murders and where they blame lie, all the way to an argument on the use of AI in warfare and the military-industrial complex.

At the end of the trial, which had been turned into a circus for the media, the final verdict was the disbanding of NEOnix, the arrest of its president and the illegalization of use of AI in military constructs without dozens of safety tests, checks and constant oversight by several committees (though ship AI and drones managed to escape this through several loopholes.)

However, due to an oversight (and possibly bribes), private companies were never added to the law. Thus, the Chevalier Errant was made by Blacklist Private Military Contractors in 2506, which began an arms race for military robots with several other PMC groups, which Blacklight came on top for most of. The public panicked at the possibility of another NEOnix Incident and UN law makers scrambled to attempt to slow the buildup of these constructs.

Despite numerous bribes and other dealings, the arms race would eventually be slowed down, especially with an amendment to the AI Accords which made it near impossible for private companies to gain smart AI constructs. By 2525, the arms race slowed to a stop as the numerous disadvantages in robot soldiers, along with their cost and their new lack of intelligence due to the amendment caused for their popularity to fall rapidly.

They still however kept a presence in Blacklight, which still used them in counter-insurgency operations in 2525, along with several smart AI models still in use.